1 00:00:05,279 --> 00:00:07,799 there we go that's me I am a freelance 2 00:00:07,799 --> 00:00:09,200 developer Advocate and one of my 3 00:00:09,200 --> 00:00:11,360 previous jobs was at a cloud company a 4 00:00:11,360 --> 00:00:12,880 cloud provider uh not one of the big 5 00:00:12,880 --> 00:00:15,960 ones but we dealt quite a bit with a lot 6 00:00:15,960 --> 00:00:21,000 of data you know um Code Media whatever 7 00:00:21,000 --> 00:00:23,039 uh and I started learning about how and 8 00:00:23,039 --> 00:00:25,000 why we store data in the way that we do 9 00:00:25,000 --> 00:00:26,840 as an industry how we did it at my 10 00:00:26,840 --> 00:00:28,720 particular employer at the time and I 11 00:00:28,720 --> 00:00:30,400 went down a lot of lot of rabbit holes 12 00:00:30,400 --> 00:00:33,480 and I came out with this talk so today 13 00:00:33,480 --> 00:00:35,800 we are going to be talking about data 14 00:00:35,800 --> 00:00:38,239 storage U before we get properly started 15 00:00:38,239 --> 00:00:40,600 it's worth asking what is this concept 16 00:00:40,600 --> 00:00:43,000 that I'm going to be talking about today 17 00:00:43,000 --> 00:00:44,680 uh one of the ways I like to describe it 18 00:00:44,680 --> 00:00:47,160 is in the Sherlock Holmes Stories the 19 00:00:47,160 --> 00:00:48,960 titular detective says that he doesn't 20 00:00:48,960 --> 00:00:50,760 know that the Earth revolves around the 21 00:00:50,760 --> 00:00:53,160 Sun and he defends this lack of 22 00:00:53,160 --> 00:00:55,079 knowledge by saying that like every 23 00:00:55,079 --> 00:00:56,680 piece of knowledge he has is like a 24 00:00:56,680 --> 00:00:58,480 piece of luggage taking up space in an 25 00:00:58,480 --> 00:01:00,440 attic his brain is like an attic 26 00:01:00,440 --> 00:01:02,879 crammed full of of knowledge and if he 27 00:01:02,879 --> 00:01:05,159 wanted to keep knowledge uh about the 28 00:01:05,159 --> 00:01:06,600 Sun going around the earth going around 29 00:01:06,600 --> 00:01:08,520 the sun he would have to kick out 30 00:01:08,520 --> 00:01:11,400 something that was more useful obviously 31 00:01:11,400 --> 00:01:13,119 like our brains don't work quite like 32 00:01:13,119 --> 00:01:15,600 that necessarily um but the general 33 00:01:15,600 --> 00:01:17,759 principle stands we need to know more 34 00:01:17,759 --> 00:01:19,439 things in our day-to-day life especially 35 00:01:19,439 --> 00:01:20,400 as the world gets more and more 36 00:01:20,400 --> 00:01:23,799 complicated than we can fit in our heads 37 00:01:23,799 --> 00:01:25,799 so for example how many people in this 38 00:01:25,799 --> 00:01:27,520 room would say they know three phone 39 00:01:27,520 --> 00:01:31,109 numbers by heart 40 00:01:31,119 --> 00:01:35,560 four five right um we need to know more 41 00:01:35,560 --> 00:01:37,320 things than we can remember and that 42 00:01:37,320 --> 00:01:39,119 means we need to store them somewhere 43 00:01:39,119 --> 00:01:41,159 but you can't store ideas you can only 44 00:01:41,159 --> 00:01:43,640 store objects so in order to store 45 00:01:43,640 --> 00:01:45,320 information we have to turn it into a 46 00:01:45,320 --> 00:01:47,799 physical object this talk is about how 47 00:01:47,799 --> 00:01:49,799 we as a species have done that 48 00:01:49,799 --> 00:01:52,000 historically how computers have impacted 49 00:01:52,000 --> 00:01:53,159 the ways that we do that and the 50 00:01:53,159 --> 00:01:56,000 problems that we went into doing that um 51 00:01:56,000 --> 00:01:57,640 in terms of computers I'm going to talk 52 00:01:57,640 --> 00:02:01,079 specifically about nonv memory so memory 53 00:02:01,079 --> 00:02:02,479 that doesn't lose its state when it 54 00:02:02,479 --> 00:02:04,000 loses its power so we're not talking 55 00:02:04,000 --> 00:02:05,640 about for example 56 00:02:05,640 --> 00:02:08,080 Ram one thing I'm also going to touch on 57 00:02:08,080 --> 00:02:10,520 is the durability of data storage which 58 00:02:10,520 --> 00:02:12,599 has two aspects to it two kind of 59 00:02:12,599 --> 00:02:15,440 flavors obviously when you turn ideas 60 00:02:15,440 --> 00:02:17,640 into objects the physical durability of 61 00:02:17,640 --> 00:02:19,599 that object matters you know if you can 62 00:02:19,599 --> 00:02:21,440 tear up a piece of paper or you know 63 00:02:21,440 --> 00:02:24,360 snap a CD in half that's part of what 64 00:02:24,360 --> 00:02:26,800 makes data durable or not durable but 65 00:02:26,800 --> 00:02:30,040 the other half is about um the encoding 66 00:02:30,040 --> 00:02:32,400 that we use the way that information is 67 00:02:32,400 --> 00:02:33,840 transformed when we turn it into a 68 00:02:33,840 --> 00:02:36,000 physical object if we lose the ability 69 00:02:36,000 --> 00:02:38,360 to transform it back and forth we also 70 00:02:38,360 --> 00:02:41,400 lose the data so who can tell me without 71 00:02:41,400 --> 00:02:45,640 getting out their phones where this 72 00:02:45,640 --> 00:02:48,280 leads so that's not very durable right 73 00:02:48,280 --> 00:02:49,800 if you don't have your phone you can't 74 00:02:49,800 --> 00:02:51,360 interpret this 75 00:02:51,360 --> 00:02:53,879 data um so that's one of the things that 76 00:02:53,879 --> 00:02:55,440 we need to be careful of when we talk 77 00:02:55,440 --> 00:02:57,440 about storing data is not just how 78 00:02:57,440 --> 00:02:58,760 what's the object that we're turning it 79 00:02:58,760 --> 00:03:01,760 into but how we do that transformation 80 00:03:01,760 --> 00:03:04,799 so what I'm going to talk about today is 81 00:03:04,799 --> 00:03:06,959 um I'm going to go through old ways of 82 00:03:06,959 --> 00:03:09,080 storing data so this is you no 83 00:03:09,080 --> 00:03:10,959 electricity before we sort of harnessed 84 00:03:10,959 --> 00:03:13,640 electricity how did we store data we've 85 00:03:13,640 --> 00:03:16,239 got old things which did use electricity 86 00:03:16,239 --> 00:03:18,560 so a l of the fun history of computing 87 00:03:18,560 --> 00:03:19,959 we've got things that we use and know 88 00:03:19,959 --> 00:03:22,560 and love and maybe hate today and then 89 00:03:22,560 --> 00:03:24,840 for a bit of fun I'm going to talk about 90 00:03:24,840 --> 00:03:27,959 things that we might use in the future 91 00:03:27,959 --> 00:03:30,239 um I make no promises as to the ACC of 92 00:03:30,239 --> 00:03:31,519 any of the things that I talk about in 93 00:03:31,519 --> 00:03:34,280 this section but they are very cool so 94 00:03:34,280 --> 00:03:37,000 let's get started old 95 00:03:37,000 --> 00:03:40,840 stuff first up we've got clay tablets 96 00:03:40,840 --> 00:03:42,840 these were traditionally not fired in a 97 00:03:42,840 --> 00:03:44,879 Kil but instead they were left to air 98 00:03:44,879 --> 00:03:47,080 dry and as a result we don't have a lot 99 00:03:47,080 --> 00:03:50,120 of these um these uh the one that was up 100 00:03:50,120 --> 00:03:52,519 here was actually found in the ruins of 101 00:03:52,519 --> 00:03:54,799 a guy's house that had burned down and 102 00:03:54,799 --> 00:03:56,400 the fire that burned his house down 103 00:03:56,400 --> 00:03:58,159 fired the clay tablet and that's why we 104 00:03:58,159 --> 00:04:00,799 still have it um this guy had a really 105 00:04:00,799 --> 00:04:02,280 really bad time right because his house 106 00:04:02,280 --> 00:04:05,040 burned down and as a result we now know 107 00:04:05,040 --> 00:04:06,640 what's on this clay tablet which is a 108 00:04:06,640 --> 00:04:08,480 complaint about how shitty of a copper 109 00:04:08,480 --> 00:04:11,200 Merchant he was so you heard of Ian air 110 00:04:11,200 --> 00:04:14,079 before yeah this is the Ian air tablet 111 00:04:14,079 --> 00:04:16,280 um it is probably it's it's the oldest 112 00:04:16,280 --> 00:04:19,639 known written customer complaint um it's 113 00:04:19,639 --> 00:04:21,239 about this this guy's house that burned 114 00:04:21,239 --> 00:04:22,639 down and it's one of the few that were 115 00:04:22,639 --> 00:04:25,720 found in his house so the writing on 116 00:04:25,720 --> 00:04:28,440 this tablet is aavan C form for a long 117 00:04:28,440 --> 00:04:30,880 time we could not read this linguists in 118 00:04:30,880 --> 00:04:32,639 the 18th and 19th century all got 119 00:04:32,639 --> 00:04:34,320 together they went well this is clearly 120 00:04:34,320 --> 00:04:36,840 language we'd like to be able to read it 121 00:04:36,840 --> 00:04:38,600 and they got together and they sort of 122 00:04:38,600 --> 00:04:41,240 reverse engineered the Acadian C form 123 00:04:41,240 --> 00:04:42,680 that's on this tablet so now we can 124 00:04:42,680 --> 00:04:45,000 decode it but for a long time that 125 00:04:45,000 --> 00:04:48,440 decode that encipherment was lost so 126 00:04:48,440 --> 00:04:50,919 while the object here is very durable 127 00:04:50,919 --> 00:04:52,360 physically although that's only by 128 00:04:52,360 --> 00:04:54,919 Chance the encoding of the data on it 129 00:04:54,919 --> 00:04:56,680 wasn't durable we had we lost it for a 130 00:04:56,680 --> 00:05:01,199 long long time um yeah tablets uh when I 131 00:05:01,199 --> 00:05:02,560 said history of data storage I meant 132 00:05:02,560 --> 00:05:04,720 like we're gone back to the BR Bronze 133 00:05:04,720 --> 00:05:06,880 Age another thing that I'm going to talk 134 00:05:06,880 --> 00:05:09,360 about is uh on the other end of the 135 00:05:09,360 --> 00:05:11,240 physical durability Spectrum we've got 136 00:05:11,240 --> 00:05:15,039 pen and paper um ink and parchment um 137 00:05:15,039 --> 00:05:16,960 you know pigments on Papyrus you know so 138 00:05:16,960 --> 00:05:19,720 on again the encoding here like with the 139 00:05:19,720 --> 00:05:21,759 clay tablet is written language so to be 140 00:05:21,759 --> 00:05:23,479 able to get data out of this thing you 141 00:05:23,479 --> 00:05:26,600 have to know the language um the photo 142 00:05:26,600 --> 00:05:28,840 of this photo is the first two pages of 143 00:05:28,840 --> 00:05:31,840 the baywolf manuscript and we don't 144 00:05:31,840 --> 00:05:33,840 actually know when this was written 145 00:05:33,840 --> 00:05:39,000 sometime between 925 sorry 975 and 10 um 146 00:05:39,000 --> 00:05:43,440 1,25 CE so about 3,000 years ago no 147 00:05:43,440 --> 00:05:46,800 sorry 1,000 years ago sorry um yeah we 148 00:05:46,800 --> 00:05:47,800 don't know when the poem itself was 149 00:05:47,800 --> 00:05:50,120 originally composed this is the only 150 00:05:50,120 --> 00:05:51,520 copy of it that we have now this is a 151 00:05:51,520 --> 00:05:53,560 foundational text in the study of Old 152 00:05:53,560 --> 00:05:56,880 English um and we only have the one copy 153 00:05:56,880 --> 00:05:57,560 of 154 00:05:57,560 --> 00:05:59,960 it this manuscript form part part of a 155 00:05:59,960 --> 00:06:01,600 library called The Cotton library after 156 00:06:01,600 --> 00:06:03,360 a guy called cotton who had a fondness 157 00:06:03,360 --> 00:06:05,919 for old and weird stuff like me and for 158 00:06:05,919 --> 00:06:08,120 a time this cotton library was stored in 159 00:06:08,120 --> 00:06:10,880 a building called Ash Burnham house does 160 00:06:10,880 --> 00:06:12,440 anyone want to guess what happened to 161 00:06:12,440 --> 00:06:14,360 Ash Burnham 162 00:06:14,360 --> 00:06:18,280 house like in Air's house it caught fire 163 00:06:18,280 --> 00:06:20,000 and unlike with clay tablets fire is 164 00:06:20,000 --> 00:06:22,639 very bad for the durability of paper and 165 00:06:22,639 --> 00:06:23,800 parchment and 166 00:06:23,800 --> 00:06:26,560 manuscripts and uh so the bearwolf 167 00:06:26,560 --> 00:06:28,599 manuscript um it survived the fire but 168 00:06:28,599 --> 00:06:30,000 it did suffer some damage and it's 169 00:06:30,000 --> 00:06:32,960 continued to acre damage since then as a 170 00:06:32,960 --> 00:06:35,720 result um so again you've got this very 171 00:06:35,720 --> 00:06:38,599 not very durable physical medium but we 172 00:06:38,599 --> 00:06:40,560 do have we can read Old English and we 173 00:06:40,560 --> 00:06:42,479 can get the data back out of it and this 174 00:06:42,479 --> 00:06:44,319 is all very interesting and you might be 175 00:06:44,319 --> 00:06:45,800 thinking you know this is very cool but 176 00:06:45,800 --> 00:06:47,880 this is supposed to be a tech talk um 177 00:06:47,880 --> 00:06:49,639 what did any of this have to do with 178 00:06:49,639 --> 00:06:52,120 technology well now I get to share one 179 00:06:52,120 --> 00:06:54,840 of my favorite facts with you the 180 00:06:54,840 --> 00:06:57,120 English word technology comes from the 181 00:06:57,120 --> 00:07:00,039 same root uh European and Proto European 182 00:07:00,039 --> 00:07:02,360 roote as the word for 183 00:07:02,360 --> 00:07:04,319 textiles historically this makes a lot 184 00:07:04,319 --> 00:07:06,360 of sense because as humans things that 185 00:07:06,360 --> 00:07:08,720 we've needed to do in the past are like 186 00:07:08,720 --> 00:07:10,319 find enough shelter so we don't die of 187 00:07:10,319 --> 00:07:13,039 exposure grow enough food to eat and 188 00:07:13,039 --> 00:07:15,319 produce enough cloth to make things like 189 00:07:15,319 --> 00:07:18,599 clothing and sales for ships and bedding 190 00:07:18,599 --> 00:07:21,080 so for a long time textiles were like 191 00:07:21,080 --> 00:07:23,280 the site of technological advancement in 192 00:07:23,280 --> 00:07:26,080 human societies for example if you spend 193 00:07:26,080 --> 00:07:28,000 time looking at Medieval manuscripts as 194 00:07:28,000 --> 00:07:30,120 I am known to do you will often see 195 00:07:30,120 --> 00:07:32,199 women with drop spindles that's what 196 00:07:32,199 --> 00:07:33,759 this woman is holding here this big long 197 00:07:33,759 --> 00:07:37,120 stick um with the sort of pointy bit at 198 00:07:37,120 --> 00:07:38,720 the top I have a laser yeah with a 199 00:07:38,720 --> 00:07:40,680 pointy bit at the top um is called a 200 00:07:40,680 --> 00:07:42,360 disf and then here she's got a drop 201 00:07:42,360 --> 00:07:43,720 spindle and you can see her cat is 202 00:07:43,720 --> 00:07:45,879 bothering her which if you do yarncrafts 203 00:07:45,879 --> 00:07:48,199 you will know is absolutely true to life 204 00:07:48,199 --> 00:07:50,000 and the thing about this illustration is 205 00:07:50,000 --> 00:07:52,479 that it's you see so many of these 206 00:07:52,479 --> 00:07:54,240 because as a woman in the you know the 207 00:07:54,240 --> 00:07:56,199 12th century if you had five spare 208 00:07:56,199 --> 00:07:57,759 minutes no you didn't you needed to 209 00:07:57,759 --> 00:08:00,960 spend it spinning um fiber from your 210 00:08:00,960 --> 00:08:03,800 sheep into yarn to be used in textiles 211 00:08:03,800 --> 00:08:04,960 because that was like the single most 212 00:08:04,960 --> 00:08:06,400 important thing you could be spending 213 00:08:06,400 --> 00:08:08,919 your time doing we get the English idiom 214 00:08:08,919 --> 00:08:10,879 to Spin A Yarn meaning to tell a story 215 00:08:10,879 --> 00:08:12,879 or to tell a tale because women would 216 00:08:12,879 --> 00:08:14,639 sit together with their drop spindles 217 00:08:14,639 --> 00:08:16,319 and their disc offs and tell each other 218 00:08:16,319 --> 00:08:18,360 stories while spinning 219 00:08:18,360 --> 00:08:19,960 yarn 220 00:08:19,960 --> 00:08:22,080 so how do we actually turn that yarn 221 00:08:22,080 --> 00:08:25,039 into cloth well this is a loom in 2nd 222 00:08:25,039 --> 00:08:27,360 Century China um they invented the draw 223 00:08:27,360 --> 00:08:28,720 Loom which is a machine for weaving 224 00:08:28,720 --> 00:08:30,360 thread into fabric which obviously 225 00:08:30,360 --> 00:08:32,839 massively sped up this process of making 226 00:08:32,839 --> 00:08:34,839 fabric which we definitely need and you 227 00:08:34,839 --> 00:08:36,240 can see here that the fabric is like 228 00:08:36,240 --> 00:08:38,000 it's not that fancy right it's got some 229 00:08:38,000 --> 00:08:41,760 stripes and we needed fabric we do do 230 00:08:41,760 --> 00:08:44,159 still need fabric um and we'll make do 231 00:08:44,159 --> 00:08:45,920 with fabric that's like plain or not 232 00:08:45,920 --> 00:08:47,839 that fancy but what we really like is 233 00:08:47,839 --> 00:08:50,680 really fancy fabric but to make that 234 00:08:50,680 --> 00:08:52,760 each of these individual threads both 235 00:08:52,760 --> 00:08:54,160 the ones that go up and down and the 236 00:08:54,160 --> 00:08:56,120 ones that are woven back and forth you 237 00:08:56,120 --> 00:08:57,959 need to like select them very carefully 238 00:08:57,959 --> 00:09:00,000 according to their color on each part of 239 00:09:00,000 --> 00:09:02,120 you know passing some threads through 240 00:09:02,120 --> 00:09:04,160 and some threads over right this is 241 00:09:04,160 --> 00:09:06,360 labor intensive and time consuming and 242 00:09:06,360 --> 00:09:08,920 very difficult to do without error if 243 00:09:08,920 --> 00:09:11,040 only there was some way we could sort of 244 00:09:11,040 --> 00:09:13,160 get that pattern information that we 245 00:09:13,160 --> 00:09:14,360 know what it's going to be based on the 246 00:09:14,360 --> 00:09:16,200 colors of the strings and like get some 247 00:09:16,200 --> 00:09:18,200 way to encode that so that the loom 248 00:09:18,200 --> 00:09:20,640 could do it automatically and this is 249 00:09:20,640 --> 00:09:22,440 where we get Punch 250 00:09:22,440 --> 00:09:24,920 Cards so how this works is as follows 251 00:09:24,920 --> 00:09:27,279 like for each little thread you what it 252 00:09:27,279 --> 00:09:29,079 is when you when you use a loom is you 253 00:09:29,079 --> 00:09:31,120 have threads that like go up like sort 254 00:09:31,120 --> 00:09:32,800 of you've got threads that go this way 255 00:09:32,800 --> 00:09:34,640 and you separate them out right and then 256 00:09:34,640 --> 00:09:36,200 you've got some other threads that go 257 00:09:36,200 --> 00:09:39,519 back and forth so this is the warp these 258 00:09:39,519 --> 00:09:41,120 are the warp threads and these are the 259 00:09:41,120 --> 00:09:43,360 weft threads and for each thread in the 260 00:09:43,360 --> 00:09:45,519 warp you want to decide which one goes 261 00:09:45,519 --> 00:09:47,440 which ones go up and which ones Go Down 262 00:09:47,440 --> 00:09:49,959 based on what color they are and so for 263 00:09:49,959 --> 00:09:51,640 each pass what you can do is you can 264 00:09:51,640 --> 00:09:52,959 have a little hook that comes down and 265 00:09:52,959 --> 00:09:54,640 goes do I want to pull this thread up or 266 00:09:54,640 --> 00:09:57,279 do I not want to pull this thread up and 267 00:09:57,279 --> 00:09:59,600 the hook can pass through the punch card 268 00:09:59,600 --> 00:10:01,399 if there is a hole there and pick up the 269 00:10:01,399 --> 00:10:03,480 thread and it cannot pass through if 270 00:10:03,480 --> 00:10:05,279 there is no hole there so depending on 271 00:10:05,279 --> 00:10:06,200 whether or not there's a hole in your 272 00:10:06,200 --> 00:10:07,760 punch C guard you will or will not 273 00:10:07,760 --> 00:10:09,680 select a thread to go into the bit of 274 00:10:09,680 --> 00:10:12,519 the warp that goes up or not ones and 275 00:10:12,519 --> 00:10:14,760 zeros and so what you have here is you 276 00:10:14,760 --> 00:10:18,640 have patent data encoded into binary and 277 00:10:18,640 --> 00:10:21,320 this is called a jackard head um a loom 278 00:10:21,320 --> 00:10:22,680 with a jackard head on it is called a 279 00:10:22,680 --> 00:10:25,160 jackard loom and they let you get fancy 280 00:10:25,160 --> 00:10:27,440 fabric with much much better Precision 281 00:10:27,440 --> 00:10:29,680 much much lower labor costs fewer 282 00:10:29,680 --> 00:10:31,240 children getting caught in the Machinery 283 00:10:31,240 --> 00:10:32,320 of all the Looms and getting like 284 00:10:32,320 --> 00:10:34,360 crushed to death which was genuinely a 285 00:10:34,360 --> 00:10:36,680 problem um and also it gave rise to 286 00:10:36,680 --> 00:10:38,839 computers but like fancy fabric fancy 287 00:10:38,839 --> 00:10:41,160 fabric here is some fancy fabric that 288 00:10:41,160 --> 00:10:43,519 was made with a jackard loom and the 289 00:10:43,519 --> 00:10:45,399 ideas behind this technology were 290 00:10:45,399 --> 00:10:47,279 developed by a few different artisans in 291 00:10:47,279 --> 00:10:50,639 the 1700s in France um ending with the 292 00:10:50,639 --> 00:10:52,279 guy called Joseph Marie jacard where it 293 00:10:52,279 --> 00:10:54,600 gets its name and the patterns from 294 00:10:54,600 --> 00:10:56,920 Punch Cards that were made in the 1700s 295 00:10:56,920 --> 00:10:59,320 are still in use today so a big win for 296 00:10:59,320 --> 00:11:01,360 data durability we know what that data 297 00:11:01,360 --> 00:11:02,959 meant and we have the objects that we 298 00:11:02,959 --> 00:11:05,279 used to encode it and we can still use 299 00:11:05,279 --> 00:11:12,190 them that's super 300 00:11:12,200 --> 00:11:16,360 cool right photography film photography 301 00:11:16,360 --> 00:11:18,120 is anyone here like a film photography 302 00:11:18,120 --> 00:11:20,639 buff like really into film photographs 303 00:11:20,639 --> 00:11:22,839 okay cool um this will be new to all of 304 00:11:22,839 --> 00:11:25,399 you then probably this is the earliest 305 00:11:25,399 --> 00:11:27,440 image derived from a photographic 306 00:11:27,440 --> 00:11:29,000 process whether or not you can call it a 307 00:11:29,000 --> 00:11:31,839 photo photograph is sort of a bit up in 308 00:11:31,839 --> 00:11:33,959 the air what's on the left is the 309 00:11:33,959 --> 00:11:36,480 earliest photographic plate and on the 310 00:11:36,480 --> 00:11:38,560 right is a colorized version of the 311 00:11:38,560 --> 00:11:40,800 resulting image and this is this is a 312 00:11:40,800 --> 00:11:42,720 building like that that's the Horizon 313 00:11:42,720 --> 00:11:44,200 there and this is the view out of a 314 00:11:44,200 --> 00:11:46,680 guy's window that he captured it was 315 00:11:46,680 --> 00:11:49,399 created by a guy called um Nissa for NES 316 00:11:49,399 --> 00:11:52,160 I'm probably mispronouncing that uh in 317 00:11:52,160 --> 00:11:54,320 1826 and what he did was he used a 318 00:11:54,320 --> 00:11:56,279 camera obscurer which is you take a 319 00:11:56,279 --> 00:11:59,279 little pinhole in a Fab in a sort of a 320 00:11:59,279 --> 00:12:01,360 surface and light filters through that 321 00:12:01,360 --> 00:12:03,560 pin hole and ends up projecting an image 322 00:12:03,560 --> 00:12:06,040 of whatever's on the other side onto a 323 00:12:06,040 --> 00:12:08,680 surface and it will be um upside 324 00:12:08,680 --> 00:12:11,160 down and the thing that he projected 325 00:12:11,160 --> 00:12:13,440 that light onto was a photosensitive 326 00:12:13,440 --> 00:12:16,000 substance um called bitamin of jir and 327 00:12:16,000 --> 00:12:17,839 it hardened differently depending on 328 00:12:17,839 --> 00:12:19,920 what kind of light hit it so what you 329 00:12:19,920 --> 00:12:22,440 end up doing is encoding that light 330 00:12:22,440 --> 00:12:24,160 information that's hitting the plate 331 00:12:24,160 --> 00:12:26,160 into texture and then you can use that 332 00:12:26,160 --> 00:12:28,079 texture to make prints that represent 333 00:12:28,079 --> 00:12:29,800 the visuals of the light 334 00:12:29,800 --> 00:12:33,160 that hit that plate um yeah I think 335 00:12:33,160 --> 00:12:35,680 that's really really cool um it's really 336 00:12:35,680 --> 00:12:37,760 interesting also as far as data storage 337 00:12:37,760 --> 00:12:40,120 goes because the encoding itself has 338 00:12:40,120 --> 00:12:42,959 very little human input it's a chemical 339 00:12:42,959 --> 00:12:45,519 process there's no language involved 340 00:12:45,519 --> 00:12:46,959 here there's not a lot of all of the 341 00:12:46,959 --> 00:12:49,240 subjectivity of Photography is in what 342 00:12:49,240 --> 00:12:51,639 you choose to capture rather or this 343 00:12:51,639 --> 00:12:52,920 kind of Photography what you choose to 344 00:12:52,920 --> 00:12:54,680 capture and not how the actual chemical 345 00:12:54,680 --> 00:12:57,600 processes occur um that's I which I 346 00:12:57,600 --> 00:12:59,839 think is that's unusual among all of the 347 00:12:59,839 --> 00:13:01,279 kinds of data storage that I'm talking 348 00:13:01,279 --> 00:13:03,279 about today pretty much apart from also 349 00:13:03,279 --> 00:13:05,959 um vinyl discs which I'm talking about 350 00:13:05,959 --> 00:13:09,959 next um but yeah film photography uh 351 00:13:09,959 --> 00:13:12,560 cool little froid this is um a dear type 352 00:13:12,560 --> 00:13:14,120 one of the slightly later forms of film 353 00:13:14,120 --> 00:13:16,320 photography first publicly available 354 00:13:16,320 --> 00:13:17,600 kind of film photography it kind of 355 00:13:17,600 --> 00:13:20,000 democratized the entire thing and this 356 00:13:20,000 --> 00:13:22,320 is very possibly the first candid photo 357 00:13:22,320 --> 00:13:25,279 ever you can see just there um that you 358 00:13:25,279 --> 00:13:26,959 can see a guy getting his shoe shined 359 00:13:26,959 --> 00:13:29,320 and the person doing it the exposure 360 00:13:29,320 --> 00:13:31,199 time for this kind of photograph or 361 00:13:31,199 --> 00:13:33,000 deera type was about four or five 362 00:13:33,000 --> 00:13:35,440 minutes and most people on this street 363 00:13:35,440 --> 00:13:37,320 were weren't staying still for that long 364 00:13:37,320 --> 00:13:38,480 right they were walking up and down and 365 00:13:38,480 --> 00:13:39,680 they don't really show up in the 366 00:13:39,680 --> 00:13:42,320 resulting image but this guy and the guy 367 00:13:42,320 --> 00:13:45,680 shining his shoe do and they probably 368 00:13:45,680 --> 00:13:47,160 never knew that they were the first 369 00:13:47,160 --> 00:13:49,320 people ever to be the subjects of a 370 00:13:49,320 --> 00:13:51,079 candid photograph which is really really 371 00:13:51,079 --> 00:13:53,480 cool obviously film photography is 372 00:13:53,480 --> 00:13:55,600 continued to progress as a discipline 373 00:13:55,600 --> 00:13:57,519 thear types like this one here was 374 00:13:57,519 --> 00:13:59,560 superseded by codian photograph s which 375 00:13:59,560 --> 00:14:01,680 are still used by hobbyists and and 376 00:14:01,680 --> 00:14:03,360 amateur well professional film 377 00:14:03,360 --> 00:14:06,360 photographers today and um you know more 378 00:14:06,360 --> 00:14:08,079 photosensitive substances have allowed 379 00:14:08,079 --> 00:14:10,800 us to keep developing this 380 00:14:10,800 --> 00:14:14,240 discipline right vinyl audio records 381 00:14:14,240 --> 00:14:17,240 here we see the the first example of a 382 00:14:17,240 --> 00:14:18,759 paradigm that's going to come up again 383 00:14:18,759 --> 00:14:21,360 and again throughout this talk which is 384 00:14:21,360 --> 00:14:23,440 you have a disc and it spins around and 385 00:14:23,440 --> 00:14:25,759 round and then you have a head you have 386 00:14:25,759 --> 00:14:28,720 like a needle here that reads data off 387 00:14:28,720 --> 00:14:31,519 that disc as it spins and you encode 388 00:14:31,519 --> 00:14:34,759 like data is sort of stored on the dis 389 00:14:34,759 --> 00:14:36,680 in concentric circles or in a very sort 390 00:14:36,680 --> 00:14:40,040 of very gradually increasing spiral so 391 00:14:40,040 --> 00:14:42,120 here what the data is is it's audio and 392 00:14:42,120 --> 00:14:44,839 it's en encoded as texture the arm goes 393 00:14:44,839 --> 00:14:47,399 up and up and up and down and um in most 394 00:14:47,399 --> 00:14:48,680 modern players what that does is it 395 00:14:48,680 --> 00:14:50,199 translates to an electrical signal that 396 00:14:50,199 --> 00:14:51,680 we then render into 397 00:14:51,680 --> 00:14:54,000 audio so the head here just a little 398 00:14:54,000 --> 00:14:56,120 needle is anyone here I should probably 399 00:14:56,120 --> 00:14:57,199 ask this about a lot of the things I 400 00:14:57,199 --> 00:15:00,320 talk about is anyone here a vinyl viny 401 00:15:00,320 --> 00:15:04,079 collector okay a handful um yeah you 402 00:15:04,079 --> 00:15:06,240 have a little needle and uh record T 403 00:15:06,240 --> 00:15:07,680 players take the signal that it produces 404 00:15:07,680 --> 00:15:09,000 and turns it into 405 00:15:09,000 --> 00:15:11,639 audio so when we come to things about 406 00:15:11,639 --> 00:15:13,680 thinking about data durability for this 407 00:15:13,680 --> 00:15:15,880 kind of storage the data here is encoded 408 00:15:15,880 --> 00:15:18,000 in the texture of the dis and the disc 409 00:15:18,000 --> 00:15:20,040 itself doesn't have any kind of 410 00:15:20,040 --> 00:15:21,839 protective layer over it it can't 411 00:15:21,839 --> 00:15:23,279 because of the way that Nal needs to 412 00:15:23,279 --> 00:15:25,519 read from it so you have to be super 413 00:15:25,519 --> 00:15:27,399 careful when handling them we actually 414 00:15:27,399 --> 00:15:29,480 get the phrase like a broken record 415 00:15:29,480 --> 00:15:30,880 from what happens if you get a scratch 416 00:15:30,880 --> 00:15:32,480 on the surface and the needle just like 417 00:15:32,480 --> 00:15:34,360 skips back and then plays the previous 418 00:15:34,360 --> 00:15:36,360 bit again that's why like a broken 419 00:15:36,360 --> 00:15:37,800 record means repetitive because that's 420 00:15:37,800 --> 00:15:39,480 what happens if you get a scratch on the 421 00:15:39,480 --> 00:15:41,079 surface of one of these or what can 422 00:15:41,079 --> 00:15:43,519 happen um physically they're pretty 423 00:15:43,519 --> 00:15:45,079 fragile and they need pretty specialized 424 00:15:45,079 --> 00:15:47,680 equipment to play so in terms of data 425 00:15:47,680 --> 00:15:50,319 durability they're pretty fragile but 426 00:15:50,319 --> 00:15:53,160 also they offer an incredible sort of um 427 00:15:53,160 --> 00:15:55,199 high quality which is why they have 428 00:15:55,199 --> 00:15:57,199 persisted as this sort of people collect 429 00:15:57,199 --> 00:15:58,519 them and listen to them even though 430 00:15:58,519 --> 00:15:59,759 they're acquire very specialist 431 00:15:59,759 --> 00:16:01,440 equipment and they're quite difficult to 432 00:16:01,440 --> 00:16:03,199 take care of 433 00:16:03,199 --> 00:16:06,680 physically right we are now into the 434 00:16:06,680 --> 00:16:10,399 wonderful world of computers electricity 435 00:16:10,399 --> 00:16:12,240 magnetism and we're going to go back to 436 00:16:12,240 --> 00:16:15,480 punch cards so Punch Cards round two uh 437 00:16:15,480 --> 00:16:18,160 a guy called Herman holth in the he 438 00:16:18,160 --> 00:16:19,920 developed machines for tabulating census 439 00:16:19,920 --> 00:16:23,560 data for the US 1890 census all that 440 00:16:23,560 --> 00:16:25,639 data was stored on Punch Cards and he 441 00:16:25,639 --> 00:16:27,839 later founded a company that got merged 442 00:16:27,839 --> 00:16:29,560 with a few other companies and 443 00:16:29,560 --> 00:16:31,519 eventually became the entity that we 444 00:16:31,519 --> 00:16:34,839 know and love today as IBM so IBM 445 00:16:34,839 --> 00:16:37,319 essentially popularized Punch 446 00:16:37,319 --> 00:16:39,600 Cards the concept here is with the same 447 00:16:39,600 --> 00:16:41,680 as with the loom a hole in the card 448 00:16:41,680 --> 00:16:43,399 allows some kind of connection to be 449 00:16:43,399 --> 00:16:46,160 made that triggers a process in Herman's 450 00:16:46,160 --> 00:16:49,120 um uh in Hol tabulating machines the 451 00:16:49,120 --> 00:16:51,079 connection that's made is a spring that 452 00:16:51,079 --> 00:16:52,920 drops down and dips into a pool of 453 00:16:52,920 --> 00:16:55,279 liquid mercury allowing an electrical 454 00:16:55,279 --> 00:16:57,160 connection to be made which is obviously 455 00:16:57,160 --> 00:16:59,079 very safe and normal and healthy to be 456 00:16:59,079 --> 00:17:01,360 around for long periods of time 457 00:17:01,360 --> 00:17:03,199 something you might notice about this is 458 00:17:03,199 --> 00:17:06,360 that there are 80 characters on on this 459 00:17:06,360 --> 00:17:08,520 punch card sort of goes runs from 1 to 460 00:17:08,520 --> 00:17:10,959 80 um just to do with like the physical 461 00:17:10,959 --> 00:17:12,160 size of these things it was what how 462 00:17:12,160 --> 00:17:14,319 they were manufactured and we see the 463 00:17:14,319 --> 00:17:16,439 legacy of this today in systems that 464 00:17:16,439 --> 00:17:18,000 assume that you want 80 characters in 465 00:17:18,000 --> 00:17:20,319 your terminal or maybe 79 depending on 466 00:17:20,319 --> 00:17:22,360 what kind of delimiter you're using um 467 00:17:22,360 --> 00:17:25,039 80 characters per line comes from Punch 468 00:17:25,039 --> 00:17:27,120 Cards Punch Cards were eventually 469 00:17:27,120 --> 00:17:29,720 superseded um by 470 00:17:29,720 --> 00:17:32,000 sorry uh yeah sorry they were eventually 471 00:17:32,000 --> 00:17:33,840 superseded by magnetic storage of 472 00:17:33,840 --> 00:17:35,760 various kinds in terms of long-term data 473 00:17:35,760 --> 00:17:38,120 storage um there's a photo that I meant 474 00:17:38,120 --> 00:17:40,360 to put in these slides of somebody with 475 00:17:40,360 --> 00:17:43,039 a stack of data of of Punch Cards and 476 00:17:43,039 --> 00:17:44,799 it's like their PHD thesis or something 477 00:17:44,799 --> 00:17:47,160 they a Computing student um at some 478 00:17:47,160 --> 00:17:49,039 point in I think the 70s and what 479 00:17:49,039 --> 00:17:50,400 they've done is they've got this stack 480 00:17:50,400 --> 00:17:52,559 of of cards and they've drawn a line 481 00:17:52,559 --> 00:17:54,919 that goes diagonally across them like 482 00:17:54,919 --> 00:17:57,840 you know sort of across the ends of each 483 00:17:57,840 --> 00:17:59,720 of these and it's so so that if they get 484 00:17:59,720 --> 00:18:01,520 out of order they know how to put them 485 00:18:01,520 --> 00:18:03,679 back into order because if you got these 486 00:18:03,679 --> 00:18:04,960 Punch Cards if you dropped them and they 487 00:18:04,960 --> 00:18:07,120 shuffled everywhere it's like somebody 488 00:18:07,120 --> 00:18:09,000 scrambled your code right it's uh you 489 00:18:09,000 --> 00:18:10,720 lose you need to be able to put them 490 00:18:10,720 --> 00:18:11,720 back into order and that was how they 491 00:18:11,720 --> 00:18:12,840 chose to do it which I thought was 492 00:18:12,840 --> 00:18:16,960 really cool so Punch Cards um and then 493 00:18:16,960 --> 00:18:18,720 these were used in these are not 494 00:18:18,720 --> 00:18:19,919 intrinsically electrical but they were 495 00:18:19,919 --> 00:18:23,080 used in the electrical early Computing 496 00:18:23,080 --> 00:18:25,640 context we now come on to drum memory 497 00:18:25,640 --> 00:18:28,799 we've got more spinning things and now 498 00:18:28,799 --> 00:18:30,320 the spinning things we are cooking with 499 00:18:30,320 --> 00:18:33,200 magnets magnetic storage works like this 500 00:18:33,200 --> 00:18:35,640 you have some kind of um ferromagnetic 501 00:18:35,640 --> 00:18:39,120 material that's iron Cobalt nickel uh 502 00:18:39,120 --> 00:18:41,200 certain rare earth metals um and you 503 00:18:41,200 --> 00:18:43,320 arrange that that metal those those 504 00:18:43,320 --> 00:18:45,120 pherom magnetic materials in some kind 505 00:18:45,120 --> 00:18:47,600 of a crystalline structure and then you 506 00:18:47,600 --> 00:18:51,240 can polarize that material um by using 507 00:18:51,240 --> 00:18:53,440 magnets to polarize certain areas one 508 00:18:53,440 --> 00:18:55,320 way or another to represent essentially 509 00:18:55,320 --> 00:18:57,320 binary data which way around is this 510 00:18:57,320 --> 00:19:00,360 area polarized drum memory works by 511 00:19:00,360 --> 00:19:02,400 having a drum that's just a cylinder 512 00:19:02,400 --> 00:19:05,280 coated in that magnetic material and 513 00:19:05,280 --> 00:19:07,960 heads or or just a single head that can 514 00:19:07,960 --> 00:19:09,520 read or write that material so it can 515 00:19:09,520 --> 00:19:11,320 either detect the polarity or change the 516 00:19:11,320 --> 00:19:14,320 polarity of the material at that point 517 00:19:14,320 --> 00:19:16,240 data is stored in tracks which are like 518 00:19:16,240 --> 00:19:19,039 circles around the surface of the drum 519 00:19:19,039 --> 00:19:21,600 and uh various different types of drum 520 00:19:21,600 --> 00:19:23,320 memory at various points have done so 521 00:19:23,320 --> 00:19:25,320 here you can see actually I don't know 522 00:19:25,320 --> 00:19:27,559 how many heads this has um but this is 523 00:19:27,559 --> 00:19:30,559 the first one this I I think is 524 00:19:30,559 --> 00:19:33,280 um yeah the best this is like the one of 525 00:19:33,280 --> 00:19:35,960 the first bits of drum memory ever made 526 00:19:35,960 --> 00:19:38,880 um but this one is called the Zam 41 527 00:19:38,880 --> 00:19:40,720 this is a Polish one and the thing 528 00:19:40,720 --> 00:19:42,080 that's interesting about this one is it 529 00:19:42,080 --> 00:19:45,679 has one read right head for every track 530 00:19:45,679 --> 00:19:47,000 because when you're dealing with your 531 00:19:47,000 --> 00:19:48,520 spinning discs and heads that read and 532 00:19:48,520 --> 00:19:51,080 write to them sometimes if your if your 533 00:19:51,080 --> 00:19:52,280 surface is very large and you've just 534 00:19:52,280 --> 00:19:54,280 got one head it'll take a long time to 535 00:19:54,280 --> 00:19:55,679 physically move the head into the 536 00:19:55,679 --> 00:19:57,120 position that it needs to be in order to 537 00:19:57,120 --> 00:19:59,280 read data and this time is called the 538 00:19:59,280 --> 00:20:01,440 seek time and you want generally to 539 00:20:01,440 --> 00:20:02,840 minimize that because especially if 540 00:20:02,840 --> 00:20:04,320 you're using these for like the working 541 00:20:04,320 --> 00:20:06,360 memory of your computer that matters 542 00:20:06,360 --> 00:20:10,159 like a lot but um if you want the one 543 00:20:10,159 --> 00:20:12,320 way to minimize that is to have multiple 544 00:20:12,320 --> 00:20:14,159 read write heads which is what this does 545 00:20:14,159 --> 00:20:16,799 it has one per track so for every layer 546 00:20:16,799 --> 00:20:18,480 that you're sort of storing data on your 547 00:20:18,480 --> 00:20:20,120 drum there's a read right head 548 00:20:20,120 --> 00:20:22,200 specifically dedicated to that one track 549 00:20:22,200 --> 00:20:25,120 but it's really expensive to do um and 550 00:20:25,120 --> 00:20:26,679 not many of these got made and they were 551 00:20:26,679 --> 00:20:28,080 not did not become popular even though 552 00:20:28,080 --> 00:20:30,320 they drastically reduced Computing time 553 00:20:30,320 --> 00:20:31,799 um because it was just really expensive 554 00:20:31,799 --> 00:20:34,240 to to build and maintain 555 00:20:34,240 --> 00:20:36,640 these we now come on to Magnetic core 556 00:20:36,640 --> 00:20:39,000 memory this was the dominant form of 557 00:20:39,000 --> 00:20:41,640 random access memory storage for decades 558 00:20:41,640 --> 00:20:43,520 and it was so ubiquitous that was often 559 00:20:43,520 --> 00:20:46,760 just referred to as core memory or core 560 00:20:46,760 --> 00:20:48,880 and the way that this works is it's also 561 00:20:48,880 --> 00:20:51,000 magnets these ones don't spin though 562 00:20:51,000 --> 00:20:53,960 each of these Rings is made of magnetic 563 00:20:53,960 --> 00:20:55,520 material and what you've got running 564 00:20:55,520 --> 00:20:58,600 through them is copper wire so generally 565 00:20:58,600 --> 00:21:00,760 at least three bits of copper wire per 566 00:21:00,760 --> 00:21:03,240 ring and by running currents through 567 00:21:03,240 --> 00:21:05,280 those copper wires you can change the 568 00:21:05,280 --> 00:21:07,799 magnet magnetism of each of those rings 569 00:21:07,799 --> 00:21:09,600 so each of those Rings represents a 570 00:21:09,600 --> 00:21:13,200 little bit um reading from a core also 571 00:21:13,200 --> 00:21:15,080 requires running a current through or 572 00:21:15,080 --> 00:21:17,440 generating a current um and what that 573 00:21:17,440 --> 00:21:19,880 does is it wipes the magnetism from that 574 00:21:19,880 --> 00:21:21,919 ring what that means is that when you 575 00:21:21,919 --> 00:21:24,480 read data from memories that stored this 576 00:21:24,480 --> 00:21:27,159 way you also destroy that data it's 577 00:21:27,159 --> 00:21:30,360 destructive readout so you can't copy 578 00:21:30,360 --> 00:21:32,880 data in this way what you have to do is 579 00:21:32,880 --> 00:21:34,279 you have to read it out and then maybe 580 00:21:34,279 --> 00:21:36,200 store it in two 581 00:21:36,200 --> 00:21:39,039 places 582 00:21:39,039 --> 00:21:41,679 um oh yeah one NE thing about this kind 583 00:21:41,679 --> 00:21:44,480 of memory is that it was heated rather 584 00:21:44,480 --> 00:21:46,240 than cooled in order to make it more 585 00:21:46,240 --> 00:21:48,279 efficient the thing that makes this kind 586 00:21:48,279 --> 00:21:50,080 of memory efficient is having a 587 00:21:50,080 --> 00:21:52,159 consistent temperature and it's a lot 588 00:21:52,159 --> 00:21:53,520 easier to manage that by heating 589 00:21:53,520 --> 00:21:56,279 something up than by cooling it down so 590 00:21:56,279 --> 00:21:57,799 if you are having you know problems with 591 00:21:57,799 --> 00:21:59,559 your PC at home and it's getting too hot 592 00:21:59,559 --> 00:22:01,159 and you maybe install some core memory 593 00:22:01,159 --> 00:22:02,559 in it or just take advantage of all of 594 00:22:02,559 --> 00:22:05,279 that heat using smaller and smaller 595 00:22:05,279 --> 00:22:09,520 cores and wires um basically lets you 596 00:22:09,520 --> 00:22:11,159 much more densely store information in 597 00:22:11,159 --> 00:22:13,320 the same physical space but 598 00:22:13,320 --> 00:22:15,080 manufacturing that as you get smaller 599 00:22:15,080 --> 00:22:17,000 and smaller becomes very cost intensive 600 00:22:17,000 --> 00:22:18,919 very labor intensive and this is a 601 00:22:18,919 --> 00:22:21,279 process that had to be done by hand so 602 00:22:21,279 --> 00:22:23,480 for working memory of your computer this 603 00:22:23,480 --> 00:22:25,760 was eventually superseded by Ram which 604 00:22:25,760 --> 00:22:27,240 is a volatile memory and I won't be 605 00:22:27,240 --> 00:22:29,440 covering it but it's also very Qui cool 606 00:22:29,440 --> 00:22:31,400 we get the phrase core dump from this 607 00:22:31,400 --> 00:22:33,240 kind of memory because this kind of 608 00:22:33,240 --> 00:22:36,000 memory does not lose its data when it 609 00:22:36,000 --> 00:22:38,880 loses power this is not volatile memory 610 00:22:38,880 --> 00:22:41,400 and so if your computer had an issue um 611 00:22:41,400 --> 00:22:43,400 you could dump like this the working 612 00:22:43,400 --> 00:22:45,320 memory the working state of the computer 613 00:22:45,320 --> 00:22:47,279 at the time that it had that issue would 614 00:22:47,279 --> 00:22:49,080 still be in here it would be dumped in 615 00:22:49,080 --> 00:22:50,720 this core that's where you get a Core 616 00:22:50,720 --> 00:22:51,960 dump 617 00:22:51,960 --> 00:22:54,279 from this is one of my my favorite 618 00:22:54,279 --> 00:22:56,360 things in the whole talk um this is this 619 00:22:56,360 --> 00:22:58,520 is memory this is well this is memory 620 00:22:58,520 --> 00:23:00,400 but this is actually a program this is 621 00:23:00,400 --> 00:23:02,840 actually a a computer program this is 622 00:23:02,840 --> 00:23:05,200 software and it's core rot memory and 623 00:23:05,200 --> 00:23:08,080 it's readon memory and what happens here 624 00:23:08,080 --> 00:23:09,919 is the process of writing to this memory 625 00:23:09,919 --> 00:23:12,640 is done during its construction how you 626 00:23:12,640 --> 00:23:14,880 thread that copper wire through these 627 00:23:14,880 --> 00:23:19,480 Rings en codes information um yeah which 628 00:23:19,480 --> 00:23:21,279 way around sort of ones and zeros you 629 00:23:21,279 --> 00:23:23,559 can kind of see how that would map um to 630 00:23:23,559 --> 00:23:24,919 update this software you've got to do 631 00:23:24,919 --> 00:23:26,960 all kinds of complicated unthreading and 632 00:23:26,960 --> 00:23:29,279 unthreading and rewiring 633 00:23:29,279 --> 00:23:31,640 and this type of memory got the name LOL 634 00:23:31,640 --> 00:23:33,960 memory or LOL memory for little old 635 00:23:33,960 --> 00:23:35,320 ladies because the process of 636 00:23:35,320 --> 00:23:37,320 manufacturing this and writing to it 637 00:23:37,320 --> 00:23:38,760 looked like sort of some kind of 638 00:23:38,760 --> 00:23:40,480 threadcraft like knitting or crochet or 639 00:23:40,480 --> 00:23:42,200 whatever you look like little old ladies 640 00:23:42,200 --> 00:23:45,039 doing it um the most famous usage of 641 00:23:45,039 --> 00:23:47,480 this was in the Apollo missions and it 642 00:23:47,480 --> 00:23:49,760 was used for that because it's really 643 00:23:49,760 --> 00:23:52,559 hard for this to change like you you 644 00:23:52,559 --> 00:23:55,400 can't it with core with core memory for 645 00:23:55,400 --> 00:23:57,600 example you can imagine there might be 646 00:23:57,600 --> 00:23:58,720 um and something else I come on to you 647 00:23:58,720 --> 00:23:59,960 later when we talk about hard dis it's 648 00:23:59,960 --> 00:24:01,520 sort of random effects that could maybe 649 00:24:01,520 --> 00:24:03,640 flip a bit right and then suddenly you 650 00:24:03,640 --> 00:24:06,000 have an error that's really hard for 651 00:24:06,000 --> 00:24:07,520 that to happen here because it's about 652 00:24:07,520 --> 00:24:09,919 how the wire is threaded through Rings 653 00:24:09,919 --> 00:24:12,279 this is really really stable and durable 654 00:24:12,279 --> 00:24:14,400 physically once it's been constructed so 655 00:24:14,400 --> 00:24:15,919 if you want something that you know you 656 00:24:15,919 --> 00:24:18,279 don't want to do updates to like your 657 00:24:18,279 --> 00:24:20,400 software for your spacecraft once it's 658 00:24:20,400 --> 00:24:22,240 already in flight and you want to make 659 00:24:22,240 --> 00:24:24,240 sure that it stays stable this is great 660 00:24:24,240 --> 00:24:26,440 for that so that's really cool I like 661 00:24:26,440 --> 00:24:27,799 that I just like it's it's a piece of 662 00:24:27,799 --> 00:24:31,559 software and it's just there physical um 663 00:24:31,559 --> 00:24:33,440 right now we come on to something that I 664 00:24:33,440 --> 00:24:35,360 feel very sad about putting in the old 665 00:24:35,360 --> 00:24:37,679 section that is floppy 666 00:24:37,679 --> 00:24:41,200 discs in the historical section seen 667 00:24:41,200 --> 00:24:44,679 here in 8 in uh 5 and a/4 inch and 3 and 668 00:24:44,679 --> 00:24:47,520 1/2 inch sizes these were the latest and 669 00:24:47,520 --> 00:24:49,960 greatest in removable storage media from 670 00:24:49,960 --> 00:24:53,200 about the mid 1970s to the early 2000s 671 00:24:53,200 --> 00:24:55,080 and has anyone ever opened up a floppy 672 00:24:55,080 --> 00:24:57,480 disc before yeah that's why they're 673 00:24:57,480 --> 00:25:00,120 called floppy discs open them up um and 674 00:25:00,120 --> 00:25:03,200 in the inside is magnetic sort of tape 675 00:25:03,200 --> 00:25:06,600 film um sort of material and that's and 676 00:25:06,600 --> 00:25:08,120 again it's a spinning disc it's a 677 00:25:08,120 --> 00:25:09,679 spinning disc with stuff recorded on it 678 00:25:09,679 --> 00:25:10,720 and tracks and when you put it in your 679 00:25:10,720 --> 00:25:12,120 computer the disc spins and something 680 00:25:12,120 --> 00:25:15,000 reads off it um this is what like a 681 00:25:15,000 --> 00:25:16,799 naked floppy disc looks like this is the 682 00:25:16,799 --> 00:25:19,279 innard of a 3.5 in floppy disc and if 683 00:25:19,279 --> 00:25:21,120 you picked it up and shook it it goes It 684 00:25:21,120 --> 00:25:23,600 goes and it's very satisfying but it 685 00:25:23,600 --> 00:25:26,120 doesn't work again after that very 686 00:25:26,120 --> 00:25:28,440 well uh here's what here's what it looks 687 00:25:28,440 --> 00:25:29,559 like really close up and you can 688 00:25:29,559 --> 00:25:31,679 actually see the way that it curves it 689 00:25:31,679 --> 00:25:33,360 sort of Curves like this obviously much 690 00:25:33,360 --> 00:25:34,440 less extreme than I'm doing with the 691 00:25:34,440 --> 00:25:36,640 laser pointer um but yeah you can see 692 00:25:36,640 --> 00:25:39,279 the tracks this was a uh photograph 693 00:25:39,279 --> 00:25:40,799 taken with um what's called a magnetic 694 00:25:40,799 --> 00:25:42,840 field camera and it detects 695 00:25:42,840 --> 00:25:44,440 magnetization so the bright and dark 696 00:25:44,440 --> 00:25:47,120 areas are areas of magnetic polarization 697 00:25:47,120 --> 00:25:50,520 on the surface of this floppy disc um 698 00:25:50,520 --> 00:25:52,720 yeah of course this means that if you 699 00:25:52,720 --> 00:25:55,080 put a magnet near it you will destroy 700 00:25:55,080 --> 00:25:57,440 the data that is encoded using magnetism 701 00:25:57,440 --> 00:25:59,760 so don't don't do that one again one of 702 00:25:59,760 --> 00:26:02,600 my favorite images on the internet um 703 00:26:02,600 --> 00:26:04,600 one thing that's uh interesting about 704 00:26:04,600 --> 00:26:06,679 floppy discs and the spinning disc kind 705 00:26:06,679 --> 00:26:08,919 of method in general is when you think 706 00:26:08,919 --> 00:26:10,600 about the way that data is encoded on 707 00:26:10,600 --> 00:26:12,480 diss it's a bit like this here we can 708 00:26:12,480 --> 00:26:16,159 see sort of two tracks right um and I've 709 00:26:16,159 --> 00:26:18,399 sort of highlighted them in one sector 710 00:26:18,399 --> 00:26:20,919 so memory would be addressed like per 711 00:26:20,919 --> 00:26:24,159 track or per sector and then per track 712 00:26:24,159 --> 00:26:25,480 um or vice versa it doesn't really 713 00:26:25,480 --> 00:26:28,000 matter but you've got sort of the thing 714 00:26:28,000 --> 00:26:29,480 that I wanted to point out is that each 715 00:26:29,480 --> 00:26:34,159 of these lines would take the same time 716 00:26:34,159 --> 00:26:36,760 to read from as the dis spins at a 717 00:26:36,760 --> 00:26:39,919 constant rate you know the it's going to 718 00:26:39,919 --> 00:26:42,679 your the head that hovers over here or 719 00:26:42,679 --> 00:26:44,720 over here is going to be reading for the 720 00:26:44,720 --> 00:26:47,000 same amount of time as it traverses here 721 00:26:47,000 --> 00:26:49,159 or if it was traversing here and what 722 00:26:49,159 --> 00:26:51,200 that means is that when you store data 723 00:26:51,200 --> 00:26:52,640 if you want your head to be reading 724 00:26:52,640 --> 00:26:54,840 every however many micros seconds or 725 00:26:54,840 --> 00:26:57,480 whatever you get to store much data much 726 00:26:57,480 --> 00:26:59,880 more densely towards the center that you 727 00:26:59,880 --> 00:27:01,880 do on the edge because here the disc is 728 00:27:01,880 --> 00:27:05,320 moving faster because it's the track is 729 00:27:05,320 --> 00:27:07,159 longer so your data has to be much more 730 00:27:07,159 --> 00:27:09,000 spread out and this is inefficient 731 00:27:09,000 --> 00:27:10,600 you're not using the surface of the 732 00:27:10,600 --> 00:27:13,320 magnetism here as efficiently as you are 733 00:27:13,320 --> 00:27:15,640 towards the center um and if you care 734 00:27:15,640 --> 00:27:17,399 very much about like optimizing the use 735 00:27:17,399 --> 00:27:18,679 of your magnetic material you don't want 736 00:27:18,679 --> 00:27:20,360 to do this You' like to store data much 737 00:27:20,360 --> 00:27:22,399 more densely at the outside but what 738 00:27:22,399 --> 00:27:24,320 that means is that when you're reading 739 00:27:24,320 --> 00:27:26,080 from the outer edge of the disc your 740 00:27:26,080 --> 00:27:28,200 disc has to be spinning more slowly you 741 00:27:28,200 --> 00:27:30,000 can't have constant angular velocity 742 00:27:30,000 --> 00:27:31,360 you'd have to have variable angular 743 00:27:31,360 --> 00:27:34,640 velocity Apple tried to do this um they 744 00:27:34,640 --> 00:27:36,640 tried to build floppy discs that worked 745 00:27:36,640 --> 00:27:38,720 that way so that they could optimize the 746 00:27:38,720 --> 00:27:40,760 amount of storage that they put on each 747 00:27:40,760 --> 00:27:43,440 bit of the spinning disc but what this 748 00:27:43,440 --> 00:27:46,000 meant is that discs encoded written to 749 00:27:46,000 --> 00:27:48,600 in this way could not be read from in a 750 00:27:48,600 --> 00:27:50,559 floppy disc drive that was not using the 751 00:27:50,559 --> 00:27:52,720 same technology one that tried to read 752 00:27:52,720 --> 00:27:55,240 from a disc written with this variable 753 00:27:55,240 --> 00:27:57,600 angular velocity technology if you tried 754 00:27:57,600 --> 00:27:58,679 to put it in a 755 00:27:58,679 --> 00:28:01,000 uh standard constant angular velocity 756 00:28:01,000 --> 00:28:03,360 disc drive you get garbage out because 757 00:28:03,360 --> 00:28:04,760 it didn't it that's not how it was 758 00:28:04,760 --> 00:28:07,399 written the encoding of the data was not 759 00:28:07,399 --> 00:28:08,919 durable in that way it required very 760 00:28:08,919 --> 00:28:10,279 specialized technology even more 761 00:28:10,279 --> 00:28:11,600 specialized than the floppy desk drive 762 00:28:11,600 --> 00:28:13,840 already is so it was expensive to 763 00:28:13,840 --> 00:28:15,600 manufacture and to maintain and they 764 00:28:15,600 --> 00:28:18,399 never caught on and so this is the model 765 00:28:18,399 --> 00:28:20,200 that most spinning disc technology uses 766 00:28:20,200 --> 00:28:22,080 is it's just less efficient at the outer 767 00:28:22,080 --> 00:28:24,279 edge but it means that we have much more 768 00:28:24,279 --> 00:28:26,840 simple technology and Hardware in order 769 00:28:26,840 --> 00:28:28,640 to read and write these things 770 00:28:28,640 --> 00:28:30,039 because the data the encoding of the 771 00:28:30,039 --> 00:28:33,200 data being easy to you know read and 772 00:28:33,200 --> 00:28:36,320 write to matters more than the physical 773 00:28:36,320 --> 00:28:38,880 efficiency of the 774 00:28:38,880 --> 00:28:41,080 medium so before computer environments 775 00:28:41,080 --> 00:28:42,600 were regularly networked floppy discs 776 00:28:42,600 --> 00:28:44,480 were the go-to way to like share data 777 00:28:44,480 --> 00:28:46,279 between different computers and of 778 00:28:46,279 --> 00:28:47,919 course this Legacy lives on with the 779 00:28:47,919 --> 00:28:49,480 floppy discs being the global symbol for 780 00:28:49,480 --> 00:28:50,919 the safe button while I've was been 781 00:28:50,919 --> 00:28:53,640 writing this talk um back in last year 782 00:28:53,640 --> 00:28:56,360 uh before NDC Oslo I think about 13 783 00:28:56,360 --> 00:28:58,480 people came up to me looked at this 784 00:28:58,480 --> 00:28:59,880 slide and went you're talking about the 785 00:28:59,880 --> 00:29:02,000 save icon and these are like you know 786 00:29:02,000 --> 00:29:04,080 50-year-old speakers who are like doing 787 00:29:04,080 --> 00:29:06,519 it as a bit but it really is a sort of 788 00:29:06,519 --> 00:29:08,200 thing it's like that's the save icon and 789 00:29:08,200 --> 00:29:09,679 that's the the Legacy that it has 790 00:29:09,679 --> 00:29:12,240 essentially because um they are now 791 00:29:12,240 --> 00:29:14,919 Wikipedia calls them obsolescent uh in 792 00:29:14,919 --> 00:29:16,960 February 2003 Dell announced that they 793 00:29:16,960 --> 00:29:18,880 would no longer pre-install floppy disc 794 00:29:18,880 --> 00:29:21,159 drives on their Flagship home computer 795 00:29:21,159 --> 00:29:24,480 and by 2007 only 2% of Home computers or 796 00:29:24,480 --> 00:29:27,480 computer sold in stores had um built-in 797 00:29:27,480 --> 00:29:28,880 drives 798 00:29:28,880 --> 00:29:32,799 um I grew up using floppy discs for like 799 00:29:32,799 --> 00:29:34,919 up until I was about a teenager and then 800 00:29:34,919 --> 00:29:37,600 USB thumb drives came became the thing 801 00:29:37,600 --> 00:29:40,360 so I'm AG I'm sort of that that's how 802 00:29:40,360 --> 00:29:44,080 old I am so it's you know they they are 803 00:29:44,080 --> 00:29:46,360 still around you can still buy adapters 804 00:29:46,360 --> 00:29:49,840 for them um the US nuclear weapons don't 805 00:29:49,840 --> 00:29:52,679 use floppy discs anymore but this uh 806 00:29:52,679 --> 00:29:55,480 this article came out in 2019 so it's 807 00:29:55,480 --> 00:29:58,360 not it's not that uh it's not that far 808 00:29:58,360 --> 00:30:00,039 into the past that these were still 809 00:30:00,039 --> 00:30:01,559 being used in really really critical 810 00:30:01,559 --> 00:30:04,240 systems um simply because they work and 811 00:30:04,240 --> 00:30:06,519 if it ain't broke don't fix it and 812 00:30:06,519 --> 00:30:08,320 another thing about floppy discs is that 813 00:30:08,320 --> 00:30:10,320 the magnetism on them is very very 814 00:30:10,320 --> 00:30:12,320 stable over time like a floppy disc that 815 00:30:12,320 --> 00:30:14,840 you had written to 20 years ago if you 816 00:30:14,840 --> 00:30:16,360 can find a way to plug it into something 817 00:30:16,360 --> 00:30:19,000 today very likely to still be readable 818 00:30:19,000 --> 00:30:22,000 and you can still do that today I went 819 00:30:22,000 --> 00:30:24,559 on Amazon I looked for um external 820 00:30:24,559 --> 00:30:26,799 floppy desk drives you can get them I 821 00:30:26,799 --> 00:30:28,480 didn't find any that were us b c 822 00:30:28,480 --> 00:30:30,559 connected but I did find lots that were 823 00:30:30,559 --> 00:30:33,360 usba um so you know they haven't kept up 824 00:30:33,360 --> 00:30:34,840 with the late like you don't have ones 825 00:30:34,840 --> 00:30:36,279 that you can plug into your MacBook that 826 00:30:36,279 --> 00:30:39,039 only has USBC you'll need an adapter but 827 00:30:39,039 --> 00:30:41,440 you can still read data from a 3.5 in 828 00:30:41,440 --> 00:30:44,159 floppy disc if you have one yeah 829 00:30:44,159 --> 00:30:45,919 obsolescent which I think is just an 830 00:30:45,919 --> 00:30:47,080 incredible 831 00:30:47,080 --> 00:30:50,149 word 832 00:30:50,159 --> 00:30:52,480 um oh and I have floppy disc stickers 833 00:30:52,480 --> 00:30:53,960 actually after this don't let me forget 834 00:30:53,960 --> 00:30:57,440 to give you your stickers um right 835 00:30:57,440 --> 00:30:59,240 current stuff stuff that we know and 836 00:30:59,240 --> 00:31:00,799 love and use 837 00:31:00,799 --> 00:31:03,559 today first up we've got more spinning 838 00:31:03,559 --> 00:31:07,360 discs Optical discs who here has um a 839 00:31:07,360 --> 00:31:10,399 desktop to PC or a laptop with a CD 840 00:31:10,399 --> 00:31:11,320 drive in 841 00:31:11,320 --> 00:31:14,279 it okay a handful who has a gaming 842 00:31:14,279 --> 00:31:16,799 console with a disc drive in it yeah 843 00:31:16,799 --> 00:31:18,440 more right this is why they're in the 844 00:31:18,440 --> 00:31:20,399 modern section it's like they're not 845 00:31:20,399 --> 00:31:22,639 that widely used in day-to-day life the 846 00:31:22,639 --> 00:31:25,200 except for you know your PS5 might still 847 00:31:25,200 --> 00:31:27,760 have a disc drive so under this category 848 00:31:27,760 --> 00:31:29,840 I'm grouping a bunch of things together 849 00:31:29,840 --> 00:31:33,360 Optical discs include like CDs uh CD um 850 00:31:33,360 --> 00:31:37,159 laser disc writable CDs Blu-rays DVDs um 851 00:31:37,159 --> 00:31:39,039 they all work in pretty much the same 852 00:31:39,039 --> 00:31:41,639 way they all use the same Paradigm as 853 00:31:41,639 --> 00:31:43,639 vinyl and floppy discs and drum memory 854 00:31:43,639 --> 00:31:45,799 you've got a thing that spins and a head 855 00:31:45,799 --> 00:31:47,679 that reads from it or something Reading 856 00:31:47,679 --> 00:31:51,159 day toare as it spins and uh Optical 857 00:31:51,159 --> 00:31:54,679 media looks like this you've got um 858 00:31:54,679 --> 00:31:57,000 bumps and dips in some polycarbonate 859 00:31:57,000 --> 00:31:59,480 plastic that and that encodes ones and 860 00:31:59,480 --> 00:32:01,760 zeros essentially then you've got a 861 00:32:01,760 --> 00:32:05,440 reflective uh layer of usually aluminium 862 00:32:05,440 --> 00:32:07,240 but not always depending on the type of 863 00:32:07,240 --> 00:32:09,960 dis and then you've got some acrylic to 864 00:32:09,960 --> 00:32:11,919 keep it flat and keep it safe and you 865 00:32:11,919 --> 00:32:14,320 might have a label on the top and the 866 00:32:14,320 --> 00:32:17,600 way that these work is that a laser 867 00:32:17,600 --> 00:32:21,480 shines down onto this surface and how 868 00:32:21,480 --> 00:32:23,880 reflective that surface is depends on 869 00:32:23,880 --> 00:32:25,760 where it is in this kind of structure 870 00:32:25,760 --> 00:32:28,240 and texture of the bumps and the laser 871 00:32:28,240 --> 00:32:29,240 will be able to tell whether it's 872 00:32:29,240 --> 00:32:31,799 looking at a bump or a dip this is a 873 00:32:31,799 --> 00:32:34,360 detector um it's a lens from an ACA 874 00:32:34,360 --> 00:32:38,240 laptop um CD D drive I think some kind 875 00:32:38,240 --> 00:32:41,279 of optical disc drive and um yeah this 876 00:32:41,279 --> 00:32:43,200 is a laser diode paired with a laser 877 00:32:43,200 --> 00:32:44,840 detector and it's detecting the 878 00:32:44,840 --> 00:32:47,039 reflectivity of the disc surface at each 879 00:32:47,039 --> 00:32:50,639 point so CD ROMs ROM standing for readon 880 00:32:50,639 --> 00:32:53,559 memory um they were developed by Sony in 881 00:32:53,559 --> 00:32:56,639 1984 and it wasn't until 1997 that we 882 00:32:56,639 --> 00:32:58,880 got rewritable CDs 883 00:32:58,880 --> 00:33:01,720 now this is a rewritable CD and you can 884 00:33:01,720 --> 00:33:03,799 notice that it's actually got a much 885 00:33:03,799 --> 00:33:06,760 darker reflective surface than the read 886 00:33:06,760 --> 00:33:08,519 only one that was the first image in 887 00:33:08,519 --> 00:33:10,840 this section and that is because it's 888 00:33:10,840 --> 00:33:12,840 not aluminium that's providing their 889 00:33:12,840 --> 00:33:15,799 reflectivity there um it's actually uh 890 00:33:15,799 --> 00:33:18,080 it's it's an alloy it is a silver indium 891 00:33:18,080 --> 00:33:20,399 antimon torium alloy try saying that 892 00:33:20,399 --> 00:33:21,559 three times fast after the party 893 00:33:21,559 --> 00:33:24,120 tomorrow night and what this does is it 894 00:33:24,120 --> 00:33:27,159 allows that laser diode in the the the 895 00:33:27,159 --> 00:33:30,039 laptop disc Drive um to actually melt 896 00:33:30,039 --> 00:33:31,960 That Metal Coating and what happens when 897 00:33:31,960 --> 00:33:34,240 it does that is it does that is that the 898 00:33:34,240 --> 00:33:35,679 Metal Coating goes from being this kind 899 00:33:35,679 --> 00:33:37,679 of crystalline structure to being like 900 00:33:37,679 --> 00:33:40,440 this kind of goopy just mess and then 901 00:33:40,440 --> 00:33:42,200 it's you know it's it's solid but it's 902 00:33:42,200 --> 00:33:44,320 it's lost a crystalline structure that 903 00:33:44,320 --> 00:33:46,240 changes the reflectivity which is what 904 00:33:46,240 --> 00:33:48,279 the laser diode reads and so it 905 00:33:48,279 --> 00:33:49,799 interprets it in the same way that the 906 00:33:49,799 --> 00:33:52,080 bumps and dips in a CD ROM are 907 00:33:52,080 --> 00:33:54,080 interpreted so you can actually write 908 00:33:54,080 --> 00:33:56,080 with the laser diode to this disc as 909 00:33:56,080 --> 00:33:58,320 well as read from it 910 00:33:58,320 --> 00:34:00,880 however in order to undo that process of 911 00:34:00,880 --> 00:34:02,320 turning the crystalline structure into 912 00:34:02,320 --> 00:34:04,919 goop you have to heat the disc up really 913 00:34:04,919 --> 00:34:07,360 you have to heat it up quite hot and you 914 00:34:07,360 --> 00:34:09,240 can't do that to like one specific point 915 00:34:09,240 --> 00:34:10,599 at once really you have to heat the 916 00:34:10,599 --> 00:34:12,800 whole disc up so if you want to erase 917 00:34:12,800 --> 00:34:15,520 what you've written onto a rewritable CD 918 00:34:15,520 --> 00:34:17,159 um you have to you have to do it all at 919 00:34:17,159 --> 00:34:21,159 once um and that's what burning a CD 920 00:34:21,159 --> 00:34:24,560 is reading data out of an optical this 921 00:34:24,560 --> 00:34:26,879 requires it to be optically legible like 922 00:34:26,879 --> 00:34:28,399 it has to look has has to look good it 923 00:34:28,399 --> 00:34:31,119 has to be legible to that laser which 924 00:34:31,119 --> 00:34:33,200 means that if you've got grease and dust 925 00:34:33,200 --> 00:34:35,200 and whatever on the surface of these I 926 00:34:35,200 --> 00:34:37,280 mean uh has anybody here not had the 927 00:34:37,280 --> 00:34:39,119 experience of playing music on a CD and 928 00:34:39,119 --> 00:34:40,599 it's got a fingerprint on it it goes all 929 00:34:40,599 --> 00:34:42,520 weird yeah like you know exactly what 930 00:34:42,520 --> 00:34:44,200 I'm talking about unlike floppy discs 931 00:34:44,200 --> 00:34:45,639 these don't have protective cases to 932 00:34:45,639 --> 00:34:47,679 protect them from that kind of thing um 933 00:34:47,679 --> 00:34:49,119 you can also break them quite 934 00:34:49,119 --> 00:34:51,000 dramatically this is a video of the 935 00:34:51,000 --> 00:34:53,200 slow-mo guys who were some um I don't 936 00:34:53,200 --> 00:34:55,079 know if they're still going but they did 937 00:34:55,079 --> 00:34:57,000 um high-speed camera capture of just 938 00:34:57,000 --> 00:34:59,040 like weird things like shooting a bullet 939 00:34:59,040 --> 00:35:00,760 into jelly or you know doing oil 940 00:35:00,760 --> 00:35:03,119 painting or whatever they span a CD so 941 00:35:03,119 --> 00:35:05,240 fast that it just 942 00:35:05,240 --> 00:35:07,520 shattered and something to note about 943 00:35:07,520 --> 00:35:09,240 this kind of data storage and it's true 944 00:35:09,240 --> 00:35:11,400 of the spinning discs in general is that 945 00:35:11,400 --> 00:35:12,839 once something like this has happened to 946 00:35:12,839 --> 00:35:14,680 it you cannot recover the data because 947 00:35:14,680 --> 00:35:16,160 it would require like lining up all the 948 00:35:16,160 --> 00:35:18,119 tracks and like you know reconciling all 949 00:35:18,119 --> 00:35:20,160 the damage that's done at those edges 950 00:35:20,160 --> 00:35:22,520 unlike for example shattering a clay 951 00:35:22,520 --> 00:35:24,079 tablet where you could piece it back 952 00:35:24,079 --> 00:35:25,640 together and maybe be able to read 953 00:35:25,640 --> 00:35:27,880 what's been written there um so in terms 954 00:35:27,880 --> 00:35:29,320 of the physical durability of these 955 00:35:29,320 --> 00:35:31,320 media it's actually got a sort of a 956 00:35:31,320 --> 00:35:32,599 different failure mode than some of the 957 00:35:32,599 --> 00:35:33,960 more analog like you know pen and paper 958 00:35:33,960 --> 00:35:36,640 clay tablets um I also really don't envy 959 00:35:36,640 --> 00:35:37,880 them having to clean up all the little 960 00:35:37,880 --> 00:35:40,040 tiny shards of CD room that they caused 961 00:35:40,040 --> 00:35:42,079 by doing this um but it's a really cool 962 00:35:42,079 --> 00:35:44,240 video I I encourage you to check it 963 00:35:44,240 --> 00:35:47,000 out right something else that's still in 964 00:35:47,000 --> 00:35:49,720 use today which you might not realize on 965 00:35:49,720 --> 00:35:51,160 uh based off the image that I'm about to 966 00:35:51,160 --> 00:35:54,720 show you is magnetic tapes so here we 967 00:35:54,720 --> 00:35:56,520 have an example of a data storage medium 968 00:35:56,520 --> 00:35:59,720 which is not random access read unlike 969 00:35:59,720 --> 00:36:01,160 hard drives and flash memory where you 970 00:36:01,160 --> 00:36:02,880 can kind of like well flash memory 971 00:36:02,880 --> 00:36:04,160 haven't covered yet but unlike hard 972 00:36:04,160 --> 00:36:06,319 drives for example where if you want to 973 00:36:06,319 --> 00:36:09,079 read from a specific purp um part of the 974 00:36:09,079 --> 00:36:11,240 surface you can move your head over and 975 00:36:11,240 --> 00:36:14,079 it has a there's a fairly short maximum 976 00:36:14,079 --> 00:36:15,680 seek time for that right it's not going 977 00:36:15,680 --> 00:36:17,800 to take too long it's got good Random 978 00:36:17,800 --> 00:36:20,440 Access but um magnetic tapes don't 979 00:36:20,440 --> 00:36:23,160 because if you're 25% the way through 980 00:36:23,160 --> 00:36:24,599 listening to Rick asley and you want to 981 00:36:24,599 --> 00:36:26,680 go to 75% the way through you have to 982 00:36:26,680 --> 00:36:29,960 scroll through that 50% it's linear seek 983 00:36:29,960 --> 00:36:32,280 time in sort of in how far away your 984 00:36:32,280 --> 00:36:34,119 data is how far away you are from where 985 00:36:34,119 --> 00:36:36,640 you want to be um and this is true of 986 00:36:36,640 --> 00:36:38,119 any kind of data stored on magnetic 987 00:36:38,119 --> 00:36:40,640 tapes basically so we'll know how this 988 00:36:40,640 --> 00:36:42,040 goes by now the tapes are made of a 989 00:36:42,040 --> 00:36:44,240 ferromagnetic material it's polarized 990 00:36:44,240 --> 00:36:46,640 ones and zeros and the hardware that 991 00:36:46,640 --> 00:36:48,760 reads and writes to the tape is known as 992 00:36:48,760 --> 00:36:52,040 a tape drive the first magnetic tapes 993 00:36:52,040 --> 00:36:54,640 were used um to store uh record computer 994 00:36:54,640 --> 00:36:58,359 data in 1951 for the univac one also an 995 00:36:58,359 --> 00:37:00,480 IBM thing oh sorry no that wasn't an IBM 996 00:37:00,480 --> 00:37:02,200 thing but IBM quickly followed suit and 997 00:37:02,200 --> 00:37:04,000 set the standards for the industry when 998 00:37:04,000 --> 00:37:06,680 use of magnetic tapes ever since um and 999 00:37:06,680 --> 00:37:08,240 they have been in constant use since 1000 00:37:08,240 --> 00:37:09,880 then pretty much and despite the 1001 00:37:09,880 --> 00:37:12,800 drawback of that non that that slow 1002 00:37:12,800 --> 00:37:14,480 Random Access time or potentially very 1003 00:37:14,480 --> 00:37:17,200 slow Random Access time um they are very 1004 00:37:17,200 --> 00:37:19,160 durable magnetism wise I mentioned this 1005 00:37:19,160 --> 00:37:21,680 with floppy discs it's same thing um you 1006 00:37:21,680 --> 00:37:24,079 can shove these in a a cupboard for a 1007 00:37:24,079 --> 00:37:26,119 few years and nothing bad will happen to 1008 00:37:26,119 --> 00:37:28,040 the data that's stored on them they're 1009 00:37:28,040 --> 00:37:30,079 very easily removable from their 1010 00:37:30,079 --> 00:37:32,119 surroundings um you know you don't have 1011 00:37:32,119 --> 00:37:34,520 to keep them powered on you don't have 1012 00:37:34,520 --> 00:37:36,680 to keep them anywhere near the the 1013 00:37:36,680 --> 00:37:38,920 things that are used to read from them 1014 00:37:38,920 --> 00:37:40,319 um which means that also the data that's 1015 00:37:40,319 --> 00:37:41,800 stored on them is not vulnerable to 1016 00:37:41,800 --> 00:37:43,040 computational 1017 00:37:43,040 --> 00:37:46,280 attack um however if you've got data in 1018 00:37:46,280 --> 00:37:48,079 a data center that's stored on a tape 1019 00:37:48,079 --> 00:37:49,359 that is just sitting in a room and been 1020 00:37:49,359 --> 00:37:51,480 removed from the computer it's going to 1021 00:37:51,480 --> 00:37:53,640 take a while if you want that data CU 1022 00:37:53,640 --> 00:37:55,280 somebody has to physically go and get it 1023 00:37:55,280 --> 00:37:56,760 and put it in a machine and scroll to 1024 00:37:56,760 --> 00:37:59,720 the right bit of the tape so it's got 1025 00:37:59,720 --> 00:38:02,160 quite a a low sorry it's got quite a 1026 00:38:02,160 --> 00:38:05,280 high latency of access to these datas 1027 00:38:05,280 --> 00:38:07,079 when data is stored on these in for 1028 00:38:07,079 --> 00:38:09,599 example um Cloud companies store this 1029 00:38:09,599 --> 00:38:10,560 and this is one of the things that I 1030 00:38:10,560 --> 00:38:11,800 learned when I was working at the cloud 1031 00:38:11,800 --> 00:38:13,800 company that I worked for um last 1032 00:38:13,800 --> 00:38:16,400 year uh is that these are very often 1033 00:38:16,400 --> 00:38:18,000 used for Glacier storage if you've got 1034 00:38:18,000 --> 00:38:19,240 sort of blob storage that you've got 1035 00:38:19,240 --> 00:38:20,440 somewhere that you never look at and 1036 00:38:20,440 --> 00:38:22,839 it's in Cold Storage it's very there's a 1037 00:38:22,839 --> 00:38:24,440 very good chance it's stored our montic 1038 00:38:24,440 --> 00:38:27,040 tapes and slow access times then are in 1039 00:38:27,040 --> 00:38:28,240 acceptable constraint because if you're 1040 00:38:28,240 --> 00:38:30,520 using it to store legal data that you 1041 00:38:30,520 --> 00:38:32,040 need to keep for compliance reasons and 1042 00:38:32,040 --> 00:38:33,599 you don't need regularly it's fine that 1043 00:38:33,599 --> 00:38:35,000 it might take 24 hours for that to 1044 00:38:35,000 --> 00:38:36,880 become available to 1045 00:38:36,880 --> 00:38:39,319 you uh one note one last note on 1046 00:38:39,319 --> 00:38:41,440 magnetti tapes one of the most famous 1047 00:38:41,440 --> 00:38:43,119 examples of something called um the 1048 00:38:43,119 --> 00:38:45,359 digital Dark Age which is the idea that 1049 00:38:45,359 --> 00:38:47,480 if we lose access to our encodings we 1050 00:38:47,480 --> 00:38:49,000 lose all that data regardless if we 1051 00:38:49,000 --> 00:38:51,599 still have the physical objects um so 1052 00:38:51,599 --> 00:38:53,160 the failure mode of that second kind of 1053 00:38:53,160 --> 00:38:56,000 data durability that I talked about um 1054 00:38:56,000 --> 00:38:57,680 one of the most famous examples of this 1055 00:38:57,680 --> 00:39:00,920 happening is do is the NASA Mission uh 1056 00:39:00,920 --> 00:39:03,480 the Viking Lander on Mars here you can 1057 00:39:03,480 --> 00:39:05,240 see car Sega next to a model they did 1058 00:39:05,240 --> 00:39:06,920 not actually send him to Mars also 1059 00:39:06,920 --> 00:39:08,240 unfortunately that would have been 1060 00:39:08,240 --> 00:39:11,560 really cool um and this Rover um 1061 00:39:11,560 --> 00:39:13,280 recorded this Lander recorded a bunch of 1062 00:39:13,280 --> 00:39:16,960 data to Magnetic tapes um NASA got the 1063 00:39:16,960 --> 00:39:20,560 tapes back in 1976 or 7 I'm quite sure 1064 00:39:20,560 --> 00:39:21,880 um they got the tapes back and then they 1065 00:39:21,880 --> 00:39:24,480 put them in a cupboard for 10 years and 1066 00:39:24,480 --> 00:39:26,119 when they got around to trying to read 1067 00:39:26,119 --> 00:39:28,040 them they could have because all the 1068 00:39:28,040 --> 00:39:29,640 engineers who had worked on this project 1069 00:39:29,640 --> 00:39:31,839 originally had either left NASA or had 1070 00:39:31,839 --> 00:39:35,200 died and nobody knew what the ones and 1071 00:39:35,200 --> 00:39:38,200 zeros on the tapes meant and they um 1072 00:39:38,200 --> 00:39:39,599 they actually had to go and look at like 1073 00:39:39,599 --> 00:39:41,000 the physical Hardware of the Viking 1074 00:39:41,000 --> 00:39:42,280 Lander in order to try and reverse 1075 00:39:42,280 --> 00:39:44,880 engineer what it was doing in order to 1076 00:39:44,880 --> 00:39:46,920 put the data onto those tapes they lost 1077 00:39:46,920 --> 00:39:51,960 the encoding um yeah and uh if they had 1078 00:39:51,960 --> 00:39:53,079 looked at those tapes a little bit 1079 00:39:53,079 --> 00:39:55,119 earlier or if the people who had um 1080 00:39:55,119 --> 00:39:56,640 designed the encodings had written them 1081 00:39:56,640 --> 00:39:57,920 down somewhere that the people in 10 1082 00:39:57,920 --> 00:39:59,599 years time were able to find they would 1083 00:39:59,599 --> 00:40:01,640 have been able to avoid that so yeah the 1084 00:40:01,640 --> 00:40:03,319 loss of the encoding very nearly lost 1085 00:40:03,319 --> 00:40:06,000 the data that was gained from sending a 1086 00:40:06,000 --> 00:40:07,880 Rover to Mars or sending a Lander to 1087 00:40:07,880 --> 00:40:11,200 Mars um that's a lot of effort to 1088 00:40:11,200 --> 00:40:14,760 waste right more spinning discs can you 1089 00:40:14,760 --> 00:40:16,920 believe how many spinning discs we have 1090 00:40:16,920 --> 00:40:19,240 um hard drives we all use them know them 1091 00:40:19,240 --> 00:40:20,880 love them you get them belt into most 1092 00:40:20,880 --> 00:40:22,280 personal computers these days unless 1093 00:40:22,280 --> 00:40:24,640 you're an SSD only person and you can 1094 00:40:24,640 --> 00:40:26,000 get an external ones that you can fill 1095 00:40:26,000 --> 00:40:28,599 with lots of pirated media and I checked 1096 00:40:28,599 --> 00:40:30,280 the other day and it looks like about 1097 00:40:30,280 --> 00:40:32,240 like 20 terabytes is commercially 1098 00:40:32,240 --> 00:40:34,640 available now they expensive but like 1099 00:40:34,640 --> 00:40:36,960 that's a lot of movies right um 1100 00:40:36,960 --> 00:40:39,400 Wikipedia says that 26 terabytes are 1101 00:40:39,400 --> 00:40:41,359 available uh as of last year but I 1102 00:40:41,359 --> 00:40:43,839 couldn't find any on Amazon or CGA or 1103 00:40:43,839 --> 00:40:46,839 whoever so these again the same Paradigm 1104 00:40:46,839 --> 00:40:49,640 as floppy discs and Optical media this 1105 00:40:49,640 --> 00:40:52,000 metal disc um in the middle spins around 1106 00:40:52,000 --> 00:40:53,560 you have a head that reads and writes to 1107 00:40:53,560 --> 00:40:56,200 it and this is again uh stored pherom 1108 00:40:56,200 --> 00:40:57,880 magnetically this is a magnetized 1109 00:40:57,880 --> 00:41:00,240 material um but they've got cases around 1110 00:41:00,240 --> 00:41:01,760 them so you're unlikely to put a magnet 1111 00:41:01,760 --> 00:41:02,920 near enough to them that you're going to 1112 00:41:02,920 --> 00:41:05,599 damage the data that's stored on them 1113 00:41:05,599 --> 00:41:08,079 once again IBM who pioneered these and 1114 00:41:08,079 --> 00:41:09,680 the first one came out in 1115 00:41:09,680 --> 00:41:13,960 1957 and stored a whopping 3.75 of God's 1116 00:41:13,960 --> 00:41:17,440 Own megabytes of data here you can see 1117 00:41:17,440 --> 00:41:18,480 two of 1118 00:41:18,480 --> 00:41:22,319 them and these are each 3.57 3.75 1119 00:41:22,319 --> 00:41:25,440 megabytes um they had these had 52 discs 1120 00:41:25,440 --> 00:41:28,119 each holding two read write heads 1121 00:41:28,119 --> 00:41:30,240 and IBM kept iterating and improving and 1122 00:41:30,240 --> 00:41:31,839 we got to where we are 1123 00:41:31,839 --> 00:41:35,800 today um yeah um they also did so again 1124 00:41:35,800 --> 00:41:38,040 with the um on the theme of things being 1125 00:41:38,040 --> 00:41:39,720 technically better but too expensive to 1126 00:41:39,720 --> 00:41:42,640 maintain and produce IBM did toy with a 1127 00:41:42,640 --> 00:41:45,040 model called the um the yeah just the 1128 00:41:45,040 --> 00:41:48,599 IBM 2305 and that had one read WR head 1129 00:41:48,599 --> 00:41:50,520 per track of data again with tracks 1130 00:41:50,520 --> 00:41:53,000 being like those concentric circles um 1131 00:41:53,000 --> 00:41:54,920 and meant that you didn't have any seek 1132 00:41:54,920 --> 00:41:57,440 time basically basically minimal seek 1133 00:41:57,440 --> 00:41:59,920 time as good as you can get um again too 1134 00:41:59,920 --> 00:42:01,640 expensive to maintain they sto making 1135 00:42:01,640 --> 00:42:04,280 them over the 1980s hard drives went 1136 00:42:04,280 --> 00:42:06,599 from being very expensive as an add-on 1137 00:42:06,599 --> 00:42:08,480 to your to your home PC to be just the 1138 00:42:08,480 --> 00:42:09,560 standard 1139 00:42:09,560 --> 00:42:11,599 essentially here we can see again 1140 00:42:11,599 --> 00:42:13,599 magnetic field camera taking image of 1141 00:42:13,599 --> 00:42:15,000 the disc 1142 00:42:15,000 --> 00:42:17,160 surface here's one of my favorite facts 1143 00:42:17,160 --> 00:42:19,319 that I've ever learned we have got to 1144 00:42:19,319 --> 00:42:21,359 the stage where the areas that we are 1145 00:42:21,359 --> 00:42:24,400 magnetizing on this surface they've got 1146 00:42:24,400 --> 00:42:26,880 so small that they've become they start 1147 00:42:26,880 --> 00:42:29,400 to get get affected by a Quantum Effect 1148 00:42:29,400 --> 00:42:32,119 called super paramagnetism and what that 1149 00:42:32,119 --> 00:42:34,240 what that does what happens when super 1150 00:42:34,240 --> 00:42:38,200 paramagnetism happens is that um heat 1151 00:42:38,200 --> 00:42:40,359 can cause magnetized material to flip 1152 00:42:40,359 --> 00:42:42,559 its par polarity right the thing that I 1153 00:42:42,559 --> 00:42:43,839 was saying would never happen with the 1154 00:42:43,839 --> 00:42:47,640 corot memory it can happen with these we 1155 00:42:47,640 --> 00:42:49,640 it it's known to happen in magnetized 1156 00:42:49,640 --> 00:42:53,800 areas um smaller than 50 nanom so for 1157 00:42:53,800 --> 00:42:55,119 reference the thickness of a piece of 1158 00:42:55,119 --> 00:42:57,200 like copy paper like pred paper is 1159 00:42:57,200 --> 00:42:59,960 anywhere between 70,000 to 180,000 1160 00:42:59,960 --> 00:43:01,760 nanometers we're talking about areas 1161 00:43:01,760 --> 00:43:04,359 that are less than 50 here um like 1162 00:43:04,359 --> 00:43:07,599 really tiny we made the magnet so small 1163 00:43:07,599 --> 00:43:09,480 that the universe doesn't like 1164 00:43:09,480 --> 00:43:11,960 it it's so cool and the way that we 1165 00:43:11,960 --> 00:43:14,160 counter this is you actually have two of 1166 00:43:14,160 --> 00:43:16,599 these discs um next to each other 1167 00:43:16,599 --> 00:43:19,040 separated by a three atom thick layer of 1168 00:43:19,040 --> 00:43:21,480 ruthenium which is not ferromagnetic 1169 00:43:21,480 --> 00:43:24,559 it's not magnetically active and the two 1170 00:43:24,559 --> 00:43:26,720 discs sitting very very close to each 1171 00:43:26,720 --> 00:43:29,240 other are polarized oppositely and then 1172 00:43:29,240 --> 00:43:31,280 they attract to each other and reinforce 1173 00:43:31,280 --> 00:43:33,040 each other's magnetism and this 1174 00:43:33,040 --> 00:43:34,839 counteracts the effect of the super 1175 00:43:34,839 --> 00:43:36,920 paramagnetism which is just really 1176 00:43:36,920 --> 00:43:40,760 really cool um yeah and that's that's 1177 00:43:40,760 --> 00:43:45,440 hard drives so going back to this is 1178 00:43:45,440 --> 00:43:49,040 um these are some photos of a data 1179 00:43:49,040 --> 00:43:51,599 storage bunker underneath Paris this 1180 00:43:51,599 --> 00:43:53,760 belongs to my previous employer a cloud 1181 00:43:53,760 --> 00:43:55,760 company called scaleway and this is 1182 00:43:55,760 --> 00:43:57,720 their cold storage and it's not magnetic 1183 00:43:57,720 --> 00:43:59,240 tape storage they actually use hard 1184 00:43:59,240 --> 00:44:01,240 drives for this which is fairly unusual 1185 00:44:01,240 --> 00:44:03,240 this data bunker is 25 M underneath 1186 00:44:03,240 --> 00:44:04,640 Paris I don't know how close it is to 1187 00:44:04,640 --> 00:44:06,760 the catacombs um and these are like the 1188 00:44:06,760 --> 00:44:08,319 only photos I could find they keep this 1189 00:44:08,319 --> 00:44:12,480 stuff locked down um and the storage 1190 00:44:12,480 --> 00:44:14,839 mechanism they use here is they use 1191 00:44:14,839 --> 00:44:16,880 these hard drives in slightly different 1192 00:44:16,880 --> 00:44:19,119 ways than we use the ones that are on 1193 00:44:19,119 --> 00:44:22,200 say your laptop or in most data centers 1194 00:44:22,200 --> 00:44:23,680 that are using hard drives for anything 1195 00:44:23,680 --> 00:44:25,960 other than cold storage so when you 1196 00:44:25,960 --> 00:44:27,920 write data to a hard drive drive you 1197 00:44:27,920 --> 00:44:29,359 have a head that reads and a head that 1198 00:44:29,359 --> 00:44:31,720 writes and sometimes they the same head 1199 00:44:31,720 --> 00:44:34,240 and when you write to a hard drive you 1200 00:44:34,240 --> 00:44:37,119 write in like quite a wide circle like a 1201 00:44:37,119 --> 00:44:40,319 fat donut right but when you're reading 1202 00:44:40,319 --> 00:44:42,119 you don't need the entirety of the width 1203 00:44:42,119 --> 00:44:43,800 of that F fat donnut you could have a 1204 00:44:43,800 --> 00:44:45,599 fairly skinny donut that you're reading 1205 00:44:45,599 --> 00:44:47,480 from does that make sense like you the 1206 00:44:47,480 --> 00:44:48,960 width that you need in order to read 1207 00:44:48,960 --> 00:44:51,599 data is much narrower than the width 1208 00:44:51,599 --> 00:44:55,000 that is written to by the right head and 1209 00:44:55,000 --> 00:44:57,599 in everyday hard drift Drive 1210 00:44:57,599 --> 00:44:58,720 that's fine it's a little bit 1211 00:44:58,720 --> 00:45:00,960 inefficient you're writing more you're 1212 00:45:00,960 --> 00:45:02,359 writing to more of the surface than you 1213 00:45:02,359 --> 00:45:05,520 need to um but it's fine that's and 1214 00:45:05,520 --> 00:45:07,720 that's what is called um parallel 1215 00:45:07,720 --> 00:45:10,319 magnetic recording where the sort of the 1216 00:45:10,319 --> 00:45:12,280 right head and the read head just 1217 00:45:12,280 --> 00:45:14,520 operate on the same tracks and they're 1218 00:45:14,520 --> 00:45:15,480 just very 1219 00:45:15,480 --> 00:45:18,839 wide in this um cold storage facility 1220 00:45:18,839 --> 00:45:20,400 what they do instead is they do 1221 00:45:20,400 --> 00:45:21,960 something called shingled magnetic 1222 00:45:21,960 --> 00:45:24,359 recording so what happens is you start 1223 00:45:24,359 --> 00:45:25,520 in the center of the disc and you write 1224 00:45:25,520 --> 00:45:27,720 a fat donut and then you move a very 1225 00:45:27,720 --> 00:45:29,640 little bit outwards and you write 1226 00:45:29,640 --> 00:45:31,400 another Fat donut that almost entirely 1227 00:45:31,400 --> 00:45:33,559 covers the first one and leaves a thin 1228 00:45:33,559 --> 00:45:36,359 strip unedited that can still be read 1229 00:45:36,359 --> 00:45:38,040 from and then you do that and you do 1230 00:45:38,040 --> 00:45:39,079 that and you do that and it's sort of 1231 00:45:39,079 --> 00:45:42,119 like you're layering up these strips 1232 00:45:42,119 --> 00:45:44,119 like the shingles on a house and it's 1233 00:45:44,119 --> 00:45:46,160 called shingled magnetic recording and 1234 00:45:46,160 --> 00:45:47,720 what this does is it lets you be much 1235 00:45:47,720 --> 00:45:49,440 much denser with the information that 1236 00:45:49,440 --> 00:45:51,839 you store to your hard drive um much 1237 00:45:51,839 --> 00:45:54,440 much better data density but the 1238 00:45:54,440 --> 00:45:56,079 trade-off is that if you wanted to go 1239 00:45:56,079 --> 00:45:58,480 back into the middle of that and write 1240 00:45:58,480 --> 00:46:00,079 you'd suddenly be covering up a lot of 1241 00:46:00,079 --> 00:46:01,920 written tracks of data with your big 1242 00:46:01,920 --> 00:46:05,079 wide read head so you cannot do random 1243 00:46:05,079 --> 00:46:07,680 rights anymore with this but in a cold 1244 00:46:07,680 --> 00:46:09,119 storage facility that's probably an 1245 00:46:09,119 --> 00:46:10,599 acceptable trade-off because you don't 1246 00:46:10,599 --> 00:46:12,359 generally go and write to data in the 1247 00:46:12,359 --> 00:46:14,920 middle of your your cold store your cold 1248 00:46:14,920 --> 00:46:16,920 stored data you're generally writing 1249 00:46:16,920 --> 00:46:19,480 blobs all at once you know your legal 1250 00:46:19,480 --> 00:46:21,720 documents you're doing backups of video 1251 00:46:21,720 --> 00:46:24,800 security footage or something um and 1252 00:46:24,800 --> 00:46:26,559 that's an acceptable trade-off and I 1253 00:46:26,559 --> 00:46:27,599 just think that's really cool that 1254 00:46:27,599 --> 00:46:28,880 there's this way that you can move right 1255 00:46:28,880 --> 00:46:30,520 to hard drives that's only used in this 1256 00:46:30,520 --> 00:46:31,720 very specific 1257 00:46:31,720 --> 00:46:33,680 circumstances um so that's what scaleway 1258 00:46:33,680 --> 00:46:35,480 does um as far as no unless they've 1259 00:46:35,480 --> 00:46:36,880 changed it and they would not let me go 1260 00:46:36,880 --> 00:46:38,720 to the data bunker while I work there so 1261 00:46:38,720 --> 00:46:40,640 I don't have to be photos of my 1262 00:46:40,640 --> 00:46:45,240 own right flash memory last but not 1263 00:46:45,240 --> 00:46:47,200 least you have some of this in your 1264 00:46:47,200 --> 00:46:49,319 pocket probably it's in smartphones it's 1265 00:46:49,319 --> 00:46:51,520 in tablets if you've got a UB key or any 1266 00:46:51,520 --> 00:46:53,599 other kind of like hardware token that's 1267 00:46:53,599 --> 00:46:56,200 using flash memory and these work by 1268 00:46:56,200 --> 00:46:57,880 using a circuit component called a 1269 00:46:57,880 --> 00:47:01,200 mosfet the a floating gate mosfet it's 1270 00:47:01,200 --> 00:47:02,880 really complicated but essentially what 1271 00:47:02,880 --> 00:47:05,280 it does is it can hold charge or it 1272 00:47:05,280 --> 00:47:08,880 cannot hold charge one and zero um yeah 1273 00:47:08,880 --> 00:47:11,559 St it stores electrical charge and the 1274 00:47:11,559 --> 00:47:13,240 thing about these devices that made them 1275 00:47:13,240 --> 00:47:15,079 so handy for when I was a teenager and 1276 00:47:15,079 --> 00:47:16,960 writing homework and taking it school is 1277 00:47:16,960 --> 00:47:18,559 that they draw power directly from 1278 00:47:18,559 --> 00:47:20,119 whatever they're plugged into they do 1279 00:47:20,119 --> 00:47:22,720 not need their own power supply unlike 1280 00:47:22,720 --> 00:47:24,599 hard drives or well some hard drives 1281 00:47:24,599 --> 00:47:26,200 draw power from the ma machine they're 1282 00:47:26,200 --> 00:47:28,480 plugged into ones but like your big one 1283 00:47:28,480 --> 00:47:30,240 with 20 terab of pirated movies that 1284 00:47:30,240 --> 00:47:33,319 needs its own power source these don't 1285 00:47:33,319 --> 00:47:35,640 um yeah this is great because they just 1286 00:47:35,640 --> 00:47:36,800 sort of plug and 1287 00:47:36,800 --> 00:47:40,119 go however flash drives and this is ssds 1288 00:47:40,119 --> 00:47:42,160 are flash drives as well um they have a 1289 00:47:42,160 --> 00:47:43,800 few limitations for a given block of 1290 00:47:43,800 --> 00:47:46,079 storage you can really only write to it 1291 00:47:46,079 --> 00:47:48,960 all at once not in patches um they also 1292 00:47:48,960 --> 00:47:50,680 have a lifetime limit on how often they 1293 00:47:50,680 --> 00:47:52,200 can be written to because those Moss 1294 00:47:52,200 --> 00:47:53,760 fats if you fill them up and then drain 1295 00:47:53,760 --> 00:47:55,200 them and fill them up and drain them 1296 00:47:55,200 --> 00:47:56,520 they start to leak they start to they 1297 00:47:56,520 --> 00:47:59,640 start to wear out um to mitigate this 1298 00:47:59,640 --> 00:48:01,319 flash drives have something in them that 1299 00:48:01,319 --> 00:48:03,119 is called a memory controller which 1300 00:48:03,119 --> 00:48:06,200 essentially tries to spread the wear 1301 00:48:06,200 --> 00:48:08,079 across all of the sort of different bits 1302 00:48:08,079 --> 00:48:09,760 of the flash drive so that all of the 1303 00:48:09,760 --> 00:48:11,280 components give out at once rather than 1304 00:48:11,280 --> 00:48:12,920 one giving out way before everything 1305 00:48:12,920 --> 00:48:15,599 else and rendering the whole flash drive 1306 00:48:15,599 --> 00:48:19,440 unusable finally those mosfets also do 1307 00:48:19,440 --> 00:48:21,440 they do leak charge over time regardless 1308 00:48:21,440 --> 00:48:23,200 of how far through their lifetime they 1309 00:48:23,200 --> 00:48:27,680 are so unlike magnetic tapes or um 1310 00:48:27,680 --> 00:48:29,800 floppy discs if you put data onto one of 1311 00:48:29,800 --> 00:48:31,240 these and then leave it in a cupboard 1312 00:48:31,240 --> 00:48:33,119 for 10 years because you are NASA after 1313 00:48:33,119 --> 00:48:35,359 the Viking Land emission you might not 1314 00:48:35,359 --> 00:48:39,559 get usable data out at the other end um 1315 00:48:39,559 --> 00:48:42,599 yeah however if you have data on a a 1316 00:48:42,599 --> 00:48:44,200 hard drive sorry not a hard drive a 1317 00:48:44,200 --> 00:48:46,839 flash disc flash drive that you want to 1318 00:48:46,839 --> 00:48:48,400 make sure you don't lose just copy it 1319 00:48:48,400 --> 00:48:49,640 onto a new one it starts the clock all 1320 00:48:49,640 --> 00:48:51,640 over again so it's 1321 00:48:51,640 --> 00:48:53,640 fine 1322 00:48:53,640 --> 00:48:56,920 yep right we have learned all about the 1323 00:48:56,920 --> 00:48:58,480 history and current state of data 1324 00:48:58,480 --> 00:49:00,720 storage um we know a bit more about how 1325 00:49:00,720 --> 00:49:02,920 and why we use various types for various 1326 00:49:02,920 --> 00:49:05,440 purposes you know we use magnetic tapes 1327 00:49:05,440 --> 00:49:08,359 for Cold Storage um and we use hard 1328 00:49:08,359 --> 00:49:10,040 drives for every day and we use the 1329 00:49:10,040 --> 00:49:11,559 weird hard drives with the shingle mag 1330 00:49:11,559 --> 00:49:14,400 recording um if you're a weird data 1331 00:49:14,400 --> 00:49:16,400 center um all the current methods of 1332 00:49:16,400 --> 00:49:18,160 storage that I've just talked about are 1333 00:49:18,160 --> 00:49:19,520 going to get they're going to be 1334 00:49:19,520 --> 00:49:20,839 iterated on over time they're going to 1335 00:49:20,839 --> 00:49:22,599 get progressively better and better 1336 00:49:22,599 --> 00:49:24,640 we'll improve the lifetime wear of flash 1337 00:49:24,640 --> 00:49:26,480 drives and hard drives will keep getting 1338 00:49:26,480 --> 00:49:28,000 big bigger and bigger and we'll find 1339 00:49:28,000 --> 00:49:29,480 ways to make the universe okay with the 1340 00:49:29,480 --> 00:49:30,920 fact that we're making the really tiny 1341 00:49:30,920 --> 00:49:33,240 magnets but none of that is like 1342 00:49:33,240 --> 00:49:36,480 particularly cool or futuristic or sexy 1343 00:49:36,480 --> 00:49:38,359 um so let's talk about some stuff that 1344 00:49:38,359 --> 00:49:43,480 is DNA DNA is very very data dense 1345 00:49:43,480 --> 00:49:45,480 microscopically small cells um can hold 1346 00:49:45,480 --> 00:49:48,000 the entire genetic code for an entire 1347 00:49:48,000 --> 00:49:50,559 species a team at Los Alamos National 1348 00:49:50,559 --> 00:49:53,200 Laboratory has created an encoding um 1349 00:49:53,200 --> 00:49:56,000 the Adaptive DNA storage codex for 1350 00:49:56,000 --> 00:49:58,559 turning Compu comp friendly binary into 1351 00:49:58,559 --> 00:50:02,160 DNA friendly pairs of AC and G bases 1352 00:50:02,160 --> 00:50:03,880 there are other encodings out there that 1353 00:50:03,880 --> 00:50:06,000 are optimized for different things like 1354 00:50:06,000 --> 00:50:08,880 density of data storage or um error 1355 00:50:08,880 --> 00:50:11,280 correction and things like that um the 1356 00:50:11,280 --> 00:50:13,200 idea here is that you can create 1357 00:50:13,200 --> 00:50:16,319 synthetic DNA that represents your data 1358 00:50:16,319 --> 00:50:18,160 and then that DNA can be stored very 1359 00:50:18,160 --> 00:50:20,880 very durably at low temperatures for 1360 00:50:20,880 --> 00:50:22,760 decades and we can actually do this 1361 00:50:22,760 --> 00:50:24,359 today there's nothing stopping us doing 1362 00:50:24,359 --> 00:50:26,880 this today people are doing it today um 1363 00:50:26,880 --> 00:50:28,440 it's just not very efficient in terms of 1364 00:50:28,440 --> 00:50:31,400 time but in terms of data durability as 1365 00:50:31,400 --> 00:50:33,839 long as we know what those encodings are 1366 00:50:33,839 --> 00:50:35,480 that means that that data could still 1367 00:50:35,480 --> 00:50:37,640 persist for a very very long time there 1368 00:50:37,640 --> 00:50:40,280 is DNA data on the moon courtesy of the 1369 00:50:40,280 --> 00:50:43,559 be sheet Lander in April 2019 it did 1370 00:50:43,559 --> 00:50:46,400 crash so we don't know what status we 1371 00:50:46,400 --> 00:50:47,880 don't know what state that DNA data is 1372 00:50:47,880 --> 00:50:51,880 in but it had synthetic DNA data on it 1373 00:50:51,880 --> 00:50:53,839 and but if it isn't lost the arch 1374 00:50:53,839 --> 00:50:55,559 Mission Foundation who planned and 1375 00:50:55,559 --> 00:50:57,160 executed that mission 1376 00:50:57,160 --> 00:50:58,599 um they estimate it could still be 1377 00:50:58,599 --> 00:51:00,520 legible if somebody knew how to read it 1378 00:51:00,520 --> 00:51:03,319 in billions of years like billions of 1379 00:51:03,319 --> 00:51:06,000 years that's incredible that's really 1380 00:51:06,000 --> 00:51:08,240 really cool when I was first researching 1381 00:51:08,240 --> 00:51:10,319 this I thought it was talking about like 1382 00:51:10,319 --> 00:51:13,880 putting DNA into like plant DNA and then 1383 00:51:13,880 --> 00:51:15,280 it's like oh I can propagate this by 1384 00:51:15,280 --> 00:51:17,920 growing the plant that has the sort of 1385 00:51:17,920 --> 00:51:20,200 um in inserted DNA in it and apparently 1386 00:51:20,200 --> 00:51:21,559 that's not a thing and there's all kinds 1387 00:51:21,559 --> 00:51:23,119 of ethical concerns with doing that and 1388 00:51:23,119 --> 00:51:24,280 that we might bring about the end of the 1389 00:51:24,280 --> 00:51:27,200 world by creating a race of super plants 1390 00:51:27,200 --> 00:51:30,880 that I don't know but um uh that's not 1391 00:51:30,880 --> 00:51:32,160 that's not what this is this is still 1392 00:51:32,160 --> 00:51:34,160 very cool though um and there are lots 1393 00:51:34,160 --> 00:51:35,799 of different encodings being considered 1394 00:51:35,799 --> 00:51:36,960 and one of the things that people are 1395 00:51:36,960 --> 00:51:38,200 trying to consider when they're building 1396 00:51:38,200 --> 00:51:40,839 these encodings is how easy would it be 1397 00:51:40,839 --> 00:51:42,799 for somebody to reverse engineer this 1398 00:51:42,799 --> 00:51:44,319 because again if we lose that encoding 1399 00:51:44,319 --> 00:51:45,640 you lose all the data that's stored and 1400 00:51:45,640 --> 00:51:48,480 people do want to mitigate that another 1401 00:51:48,480 --> 00:51:50,119 type of data storage that being looked 1402 00:51:50,119 --> 00:51:52,760 at with an eye to data durability is 1403 00:51:52,760 --> 00:51:55,240 Superman storage also known as 5D 1404 00:51:55,240 --> 00:51:57,280 Optical data storage 1405 00:51:57,280 --> 00:51:58,880 um that's a bit of a marketing name I'll 1406 00:51:58,880 --> 00:52:01,119 talk about that in a bit um the idea 1407 00:52:01,119 --> 00:52:03,640 here is that you take a transparent 1408 00:52:03,640 --> 00:52:05,760 nonphotosensitive materials so something 1409 00:52:05,760 --> 00:52:08,559 like a bit of hardened glass or fused 1410 00:52:08,559 --> 00:52:11,559 quartz and you use a laser to engrave 1411 00:52:11,559 --> 00:52:15,839 data within the 3D sort of material that 1412 00:52:15,839 --> 00:52:18,920 you've got here it's called 5D because 1413 00:52:18,920 --> 00:52:20,960 it's it's a marketing thing but what 1414 00:52:20,960 --> 00:52:22,280 they're what they're sort of getting at 1415 00:52:22,280 --> 00:52:23,760 is like you have your three physical 1416 00:52:23,760 --> 00:52:25,200 dimensions like you know back and forth 1417 00:52:25,200 --> 00:52:27,319 up and down left to right and then you 1418 00:52:27,319 --> 00:52:29,760 have what the angle is is another degree 1419 00:52:29,760 --> 00:52:31,920 of freedom and how close you are to the 1420 00:52:31,920 --> 00:52:33,280 thing as another degree of freedom and 1421 00:52:33,280 --> 00:52:35,880 they say that's five Dimensions it's not 1422 00:52:35,880 --> 00:52:39,960 but it's a good name so um yeah so what 1423 00:52:39,960 --> 00:52:42,839 happens is that depending on how close 1424 00:52:42,839 --> 00:52:44,760 you are what angle you're looking at and 1425 00:52:44,760 --> 00:52:47,200 where in the in the crystal or material 1426 00:52:47,200 --> 00:52:48,880 you're looking you can read different 1427 00:52:48,880 --> 00:52:51,200 you can read data in different ways um 1428 00:52:51,200 --> 00:52:52,839 the photo here is the universal 1429 00:52:52,839 --> 00:52:54,640 Declaration of Human Rights engraved 1430 00:52:54,640 --> 00:52:56,119 into a piece of glass and you can see 1431 00:52:56,119 --> 00:52:57,720 they have done it in a way that it looks 1432 00:52:57,720 --> 00:52:59,520 like the UN symbol which is very cool 1433 00:52:59,520 --> 00:53:03,119 and cute um Hitachi and Microsoft are 1434 00:53:03,119 --> 00:53:05,200 both working on this concept and the 1435 00:53:05,200 --> 00:53:06,640 current bleeding edge stuff is being 1436 00:53:06,640 --> 00:53:09,839 done at Southampton here in the UK all 1437 00:53:09,839 --> 00:53:11,640 of those teams are working on this with 1438 00:53:11,640 --> 00:53:13,839 an i to data durability like the 1439 00:53:13,839 --> 00:53:16,280 explicit goal of data durability the 1440 00:53:16,280 --> 00:53:18,559 data isn't going anywhere unless you 1441 00:53:18,559 --> 00:53:20,160 drop the glass which you absolutely can 1442 00:53:20,160 --> 00:53:23,040 do but as long as the glass is not like 1443 00:53:23,040 --> 00:53:25,680 shattered this isn't this isn't charge 1444 00:53:25,680 --> 00:53:27,799 that's dissipating for mosfets this 1445 00:53:27,799 --> 00:53:29,480 isn't going to get erased by putting a 1446 00:53:29,480 --> 00:53:31,599 magnet too close to it you know this is 1447 00:53:31,599 --> 00:53:34,240 very very durable physically speaking 1448 00:53:34,240 --> 00:53:36,119 the encoding that's being used to turn 1449 00:53:36,119 --> 00:53:38,440 data into engraving isn't yet settled as 1450 00:53:38,440 --> 00:53:40,680 far as I can tell and that's the much 1451 00:53:40,680 --> 00:53:42,599 more philosophically interesting and 1452 00:53:42,599 --> 00:53:45,359 difficult problem um it maybe I I I 1453 00:53:45,359 --> 00:53:46,400 don't know what it is but maybe the 1454 00:53:46,400 --> 00:53:48,240 human rights declaration is all there in 1455 00:53:48,240 --> 00:53:50,400 just like seven bit asky or something as 1456 00:53:50,400 --> 00:53:52,440 long as we know what that is in like a 1457 00:53:52,440 --> 00:53:54,119 thousand years we'd be able to get the 1458 00:53:54,119 --> 00:53:55,520 UN Declaration of Human Rights back out 1459 00:53:55,520 --> 00:53:57,240 of this piece of glass which is really 1460 00:53:57,240 --> 00:53:58,839 really cool and it's very dense as well 1461 00:53:58,839 --> 00:54:01,559 you know you can just it's quite small 1462 00:54:01,559 --> 00:54:05,480 um yeah that's really really cool so the 1463 00:54:05,480 --> 00:54:07,280 last thing that I have to talk today 1464 00:54:07,280 --> 00:54:09,720 about today and the weirdest and the one 1465 00:54:09,720 --> 00:54:11,280 where I want to caveat very strongly 1466 00:54:11,280 --> 00:54:13,400 that um I'm not actually predicting this 1467 00:54:13,400 --> 00:54:14,720 this is going to go anywhere but it's 1468 00:54:14,720 --> 00:54:17,680 really cool is time 1469 00:54:17,680 --> 00:54:20,160 crystals do we have any condensed matter 1470 00:54:20,160 --> 00:54:21,559 physicists in the 1471 00:54:21,559 --> 00:54:24,520 room 1472 00:54:24,520 --> 00:54:26,720 good I have given this talk for when 1473 00:54:26,720 --> 00:54:27,839 people have put their hands up and I was 1474 00:54:27,839 --> 00:54:29,200 like I am so sorry for what I'm about to 1475 00:54:29,200 --> 00:54:32,040 do to your beautiful profession so this 1476 00:54:32,040 --> 00:54:34,599 up here isn't a Time Crystal there are 1477 00:54:34,599 --> 00:54:36,520 no good photos of them unfortunately it 1478 00:54:36,520 --> 00:54:39,640 is a space Crystal and like flash memory 1479 00:54:39,640 --> 00:54:41,799 you have some of this in your pocket 1480 00:54:41,799 --> 00:54:43,799 because silicon is this is silicon and 1481 00:54:43,799 --> 00:54:46,040 silicon is a space Crystal the thing 1482 00:54:46,040 --> 00:54:47,760 that makes it a space crystal is it has 1483 00:54:47,760 --> 00:54:50,880 a repeating structure in space you look 1484 00:54:50,880 --> 00:54:52,520 at it and then you move over a certain 1485 00:54:52,520 --> 00:54:53,680 bit and you look at it again and they 1486 00:54:53,680 --> 00:54:55,520 look the same a repeating structure in 1487 00:54:55,520 --> 00:54:58,559 space is a space Crystal time crystals 1488 00:54:58,559 --> 00:55:00,480 are the same but they have a repeating 1489 00:55:00,480 --> 00:55:02,160 structure in 1490 00:55:02,160 --> 00:55:04,799 time this is the best image I could kind 1491 00:55:04,799 --> 00:55:06,520 of think of to try and display the 1492 00:55:06,520 --> 00:55:09,160 concept that I'm talking about um a Time 1493 00:55:09,160 --> 00:55:11,599 crystal is an object whose lowest energy 1494 00:55:11,599 --> 00:55:14,480 state it sort of resting state is one of 1495 00:55:14,480 --> 00:55:16,920 motion this doesn't break the 1496 00:55:16,920 --> 00:55:18,640 conservation of energy because something 1497 00:55:18,640 --> 00:55:20,240 something quantum 1498 00:55:20,240 --> 00:55:22,240 physics they were first theorized to 1499 00:55:22,240 --> 00:55:24,400 exist in 2012 um by a theoretical 1500 00:55:24,400 --> 00:55:26,960 physicist called France wilchek 1501 00:55:26,960 --> 00:55:30,119 and or Frank wilchek sorry and in 2021 1502 00:55:30,119 --> 00:55:32,520 so only nine years later a team of 1503 00:55:32,520 --> 00:55:34,520 scientists in the University of Hamburg 1504 00:55:34,520 --> 00:55:37,599 made one in a lab it didn't look like 1505 00:55:37,599 --> 00:55:40,400 what you might expect um it was called 1506 00:55:40,400 --> 00:55:43,119 uh it's a Bose Einstein condensate and 1507 00:55:43,119 --> 00:55:44,640 that is a state of matter like you know 1508 00:55:44,640 --> 00:55:47,760 you've got liquid gas solid um plasma 1509 00:55:47,760 --> 00:55:49,599 and then there's like 13 others or 1510 00:55:49,599 --> 00:55:50,880 something and one of those is a Bose 1511 00:55:50,880 --> 00:55:53,160 Einstein condensate um it's what you get 1512 00:55:53,160 --> 00:55:55,799 when you super cool Bon gases I 1513 00:55:55,799 --> 00:55:57,359 personally hate when I open my freezer 1514 00:55:57,359 --> 00:55:59,039 and I've actually made a Bo accidentally 1515 00:55:59,039 --> 00:56:01,599 made Bose Einstein condensate again and 1516 00:56:01,599 --> 00:56:03,960 the reason that they're in this talk is 1517 00:56:03,960 --> 00:56:05,520 because some very clever people think 1518 00:56:05,520 --> 00:56:07,119 that we might be able to use them as 1519 00:56:07,119 --> 00:56:08,880 storage for quantum 1520 00:56:08,880 --> 00:56:11,079 computers and that's currently an open 1521 00:56:11,079 --> 00:56:13,760 problem as I understand it because in 1522 00:56:13,760 --> 00:56:15,440 quantum theory there's something called 1523 00:56:15,440 --> 00:56:17,520 the cloning theorem and what the cloning 1524 00:56:17,520 --> 00:56:19,680 theorem says is you cannot duplicate 1525 00:56:19,680 --> 00:56:22,359 Quantum State you can't copy it you can 1526 00:56:22,359 --> 00:56:24,640 only read it and by reading it you 1527 00:56:24,640 --> 00:56:26,839 destroy it like with for memory 1528 00:56:26,839 --> 00:56:30,440 destructive readout so any operation um 1529 00:56:30,440 --> 00:56:33,000 that reads from a Quantum memory store 1530 00:56:33,000 --> 00:56:34,480 is intrinsically 1531 00:56:34,480 --> 00:56:37,039 destructive um and that means you know 1532 00:56:37,039 --> 00:56:38,240 that that has certain implications for 1533 00:56:38,240 --> 00:56:39,760 what we can use in order to store 1534 00:56:39,760 --> 00:56:42,839 Quantum data um but maybe it's time 1535 00:56:42,839 --> 00:56:45,720 crystals and I read the I if you want to 1536 00:56:45,720 --> 00:56:48,319 go and read up up on time crystals it's 1537 00:56:48,319 --> 00:56:50,039 an absolute trip I think I briefly 1538 00:56:50,039 --> 00:56:52,720 understood them for about like 3 seconds 1539 00:56:52,720 --> 00:56:54,039 had an out of body experience and then 1540 00:56:54,039 --> 00:56:55,160 went back to just being like okay I 1541 00:56:55,160 --> 00:56:56,079 don't I don't know what this going on 1542 00:56:56,079 --> 00:56:56,920 here 1543 00:56:56,920 --> 00:56:59,000 um but they're really really cool and 1544 00:56:59,000 --> 00:57:01,119 the fact that we sort of went well the 1545 00:57:01,119 --> 00:57:03,400 laws of physics say that this should be 1546 00:57:03,400 --> 00:57:05,920 possible I guess and then it was only 1547 00:57:05,920 --> 00:57:08,119 from nine years from that moment that 1548 00:57:08,119 --> 00:57:10,480 people made one is incredible people are 1549 00:57:10,480 --> 00:57:12,079 incredible scientists are 1550 00:57:12,079 --> 00:57:15,599 incredible so on that note conclusions 1551 00:57:15,599 --> 00:57:17,359 we have uh we have learned about storing 1552 00:57:17,359 --> 00:57:18,480 data all the way from customer 1553 00:57:18,480 --> 00:57:20,480 complaints in the Bronze Age through to 1554 00:57:20,480 --> 00:57:22,000 encoding our data on DNA and putting it 1555 00:57:22,000 --> 00:57:24,280 on the moon um this has been an 1556 00:57:24,280 --> 00:57:26,640 incomplete and Whistle Stop tour and 1557 00:57:26,640 --> 00:57:28,200 most of what I wanted to impart with 1558 00:57:28,200 --> 00:57:29,960 this is that humans are really cool and 1559 00:57:29,960 --> 00:57:32,240 creative and we're very good at solving 1560 00:57:32,240 --> 00:57:34,039 problems in increasingly complicated 1561 00:57:34,039 --> 00:57:36,400 ways um if there's another takeaway from 1562 00:57:36,400 --> 00:57:38,480 it though it is that data is useless if 1563 00:57:38,480 --> 00:57:40,720 you can't read it data durability 1564 00:57:40,720 --> 00:57:42,960 matters when you encode data you need to 1565 00:57:42,960 --> 00:57:44,480 make sure that somebody is able to read 1566 00:57:44,480 --> 00:57:46,720 it otherwise there's no point in storing 1567 00:57:46,720 --> 00:57:49,359 it when we store data it has to be with 1568 00:57:49,359 --> 00:57:50,880 that Concept in mind and this isn't 1569 00:57:50,880 --> 00:57:52,280 something that in our day-to-day job as 1570 00:57:52,280 --> 00:57:53,960 software Engineers we get to deal with 1571 00:57:53,960 --> 00:57:56,280 directly but it does come into play 1572 00:57:56,280 --> 00:57:57,880 slightly when we think talk about things 1573 00:57:57,880 --> 00:58:00,760 like encoding Concepts into encoding 1574 00:58:00,760 --> 00:58:03,400 real world concept into code objects how 1575 00:58:03,400 --> 00:58:05,760 are we how are we serializing that is it 1576 00:58:05,760 --> 00:58:07,079 in a way that makes sense to the person 1577 00:58:07,079 --> 00:58:08,240 who's going to have to deal with it at 1578 00:58:08,240 --> 00:58:11,160 the other end um and you know we get 1579 00:58:11,160 --> 00:58:12,599 into all ideas of like demander of and 1580 00:58:12,599 --> 00:58:14,039 design when we start talking about those 1581 00:58:14,039 --> 00:58:17,079 things but it all interconnects and um 1582 00:58:17,079 --> 00:58:18,760 the way that we store data is absolutely 1583 00:58:18,760 --> 00:58:21,000 part of that too all of the data is 1584 00:58:21,000 --> 00:58:22,799 useless and another person in the future 1585 00:58:22,799 --> 00:58:25,000 which might be you knows how to get at 1586 00:58:25,000 --> 00:58:28,680 it um but mostly people are really cool 1587 00:58:28,680 --> 00:58:31,160 science is really cool time 1588 00:58:31,160 --> 00:58:33,280 crystals thank you so much for your time 1589 00:58:33,280 --> 00:58:35,799 today uh if you follow that QR code it's 1590 00:58:35,799 --> 00:58:37,839 not a Rick Roll unlike the first one 1591 00:58:37,839 --> 00:58:39,720 that will take you to a little uh 1592 00:58:39,720 --> 00:58:40,960 feedback form where you can leave 1593 00:58:40,960 --> 00:58:43,880 feedback for me if you're so inclined um 1594 00:58:43,880 --> 00:58:44,920 and if you would like to give me 1595 00:58:44,920 --> 00:58:46,200 feedback I really value it but more 1596 00:58:46,200 --> 00:58:48,039 crucially I have stickers and if anyone 1597 00:58:48,039 --> 00:58:49,400 would like stickers please come up to 1598 00:58:49,400 --> 00:58:53,950 the front and we have five minutes for 1599 00:58:53,960 --> 00:58:59,750 questions no sticker time oh 1600 00:58:59,760 --> 00:59:04,109 question 1601 00:59:04,119 --> 00:59:06,480 yeah oh good question so the question 1602 00:59:06,480 --> 00:59:09,280 was the core rope memory is that where 1603 00:59:09,280 --> 00:59:11,520 hard the term hardwire comes from 1604 00:59:11,520 --> 00:59:14,640 hardwired in I don't know I could see it 1605 00:59:14,640 --> 00:59:17,520 being from a lot of different places um 1606 00:59:17,520 --> 00:59:18,640 it would be cool though and it would 1607 00:59:18,640 --> 00:59:22,039 make sense yeah yeah I can add another 1608 00:59:22,039 --> 00:59:24,440 Linguistics uh atmology fact to my talk 1609 00:59:24,440 --> 00:59:26,400 yes 1610 00:59:26,400 --> 00:59:30,599 this uh SP uh same speed does that 1611 00:59:30,599 --> 00:59:33,079 happen with like CDs and other yes I 1612 00:59:33,079 --> 00:59:35,559 think CDs are generally um constant 1613 00:59:35,559 --> 00:59:37,680 angular velocity as well the question 1614 00:59:37,680 --> 00:59:39,559 was um spinning discs about spinning 1615 00:59:39,559 --> 00:59:41,559 discs and constant speed also have like 1616 00:59:41,559 --> 00:59:45,520 different uh densities of I believe so 1617 00:59:45,520 --> 00:59:48,280 that yeah they do they do thank 1618 00:59:48,280 --> 00:59:51,950 you 1619 00:59:51,960 --> 00:59:54,799 cool uh actually about the uh spinning 1620 00:59:54,799 --> 00:59:56,680 disc thing you were talking about how 1621 00:59:56,680 --> 00:59:59,799 Apple uh had this idea to do variable 1622 00:59:59,799 --> 01:00:01,720 angular velocities and it just occurred 1623 01:00:01,720 --> 01:00:03,880 to me that that would be a great way to 1624 01:00:03,880 --> 01:00:06,680 uh include you know exclusive content 1625 01:00:06,680 --> 01:00:09,119 only available that oh yeah yeah vendor 1626 01:00:09,119 --> 01:00:11,440 look in which is you know it gives its 1627 01:00:11,440 --> 01:00:13,319 own implications listen if anyone's 1628 01:00:13,319 --> 01:00:14,920 going to do it I think it would be apple 1629 01:00:14,920 --> 01:00:18,760 right yeah yeah no um exclusive uh you 1630 01:00:18,760 --> 01:00:20,119 know Apple only content that you can 1631 01:00:20,119 --> 01:00:21,720 only get by using their specific floppy 1632 01:00:21,720 --> 01:00:23,480 disc encoding if they're going to bring 1633 01:00:23,480 --> 01:00:24,559 back the floppy disc I wouldn't be 1634 01:00:24,559 --> 01:00:27,039 opposed right and that has its own 1635 01:00:27,039 --> 01:00:31,480 implications for um uh encod ability 1636 01:00:31,480 --> 01:00:34,079 what was the term you use like data data 1637 01:00:34,079 --> 01:00:37,039 durability data durability because if 1638 01:00:37,039 --> 01:00:39,680 you if you run around doing you know and 1639 01:00:39,680 --> 01:00:43,160 we've already had problems with DRM and 1640 01:00:43,160 --> 01:00:45,039 people can't get their own stuff so I I 1641 01:00:45,039 --> 01:00:46,760 named Dr the digital Dark Age that's a 1642 01:00:46,760 --> 01:00:48,160 whole thing you can go and research and 1643 01:00:48,160 --> 01:00:50,160 it's like a topic in its own right but 1644 01:00:50,160 --> 01:00:53,119 the concept there is that as we sort of 1645 01:00:53,119 --> 01:00:54,640 encode things in very in like 1646 01:00:54,640 --> 01:00:56,559 increasingly obscure ways 1647 01:00:56,559 --> 01:00:59,559 and do things like DRM and um things 1648 01:00:59,559 --> 01:01:02,640 like Netflix taking down movies before 1649 01:01:02,640 --> 01:01:04,039 they've even really been released and 1650 01:01:04,039 --> 01:01:05,920 just like those are never ever going to 1651 01:01:05,920 --> 01:01:09,000 be seen again or whatever like how do we 1652 01:01:09,000 --> 01:01:12,119 do digital preservation um in a way that 1653 01:01:12,119 --> 01:01:15,920 is actually going to be resistant to the 1654 01:01:15,920 --> 01:01:17,200 kind of things that go wrong over time 1655 01:01:17,200 --> 01:01:18,599 when we lose information and we lose 1656 01:01:18,599 --> 01:01:22,880 access to Hardware um yeah the digital 1657 01:01:22,880 --> 01:01:26,079 Dark Age it's it's a like scary idea but 1658 01:01:26,079 --> 01:01:27,599 lots of very clever people are working 1659 01:01:27,599 --> 01:01:29,760 on really interesting ways to try and 1660 01:01:29,760 --> 01:01:32,000 fix it circumvent it protect us against 1661 01:01:32,000 --> 01:01:34,710 it 1662 01:01:34,720 --> 01:01:37,000 yeah have two more question or two more 1663 01:01:37,000 --> 01:01:39,640 minutes for questions or we can just 1664 01:01:39,640 --> 01:01:42,440 go question is there a medium you didn't 1665 01:01:42,440 --> 01:01:44,520 have time to talk about that you want to 1666 01:01:44,520 --> 01:01:46,160 mention is there a medium I didn't have 1667 01:01:46,160 --> 01:01:49,000 time to mention so the one that I went 1668 01:01:49,000 --> 01:01:50,200 back and forth on whether I wanted to 1669 01:01:50,200 --> 01:01:53,279 put it in the talk is a um South 1670 01:01:53,279 --> 01:01:54,760 American and I can't remember which 1671 01:01:54,760 --> 01:01:57,559 South American country it is is kipo and 1672 01:01:57,559 --> 01:02:00,720 they are what a kipo is is a series of 1673 01:02:00,720 --> 01:02:02,960 it's It's a fiber 1674 01:02:02,960 --> 01:02:07,039 based writing system and it's strings or 1675 01:02:07,039 --> 01:02:09,599 like ropes that are knotted in order to 1676 01:02:09,599 --> 01:02:11,880 encode data and there are a couple 1677 01:02:11,880 --> 01:02:13,480 different kinds that people think that 1678 01:02:13,480 --> 01:02:15,119 they may have been used for like maths 1679 01:02:15,119 --> 01:02:16,640 and sort of going okay this is how much 1680 01:02:16,640 --> 01:02:18,960 grain you owe me in order for this but 1681 01:02:18,960 --> 01:02:21,000 they've also been used for narrative 1682 01:02:21,000 --> 01:02:22,960 recording and one of the things that's 1683 01:02:22,960 --> 01:02:25,359 really really cool about them is that 1684 01:02:25,359 --> 01:02:27,880 sometimes it's you need to see what the 1685 01:02:27,880 --> 01:02:29,720 fiber is in order to know data like 1686 01:02:29,720 --> 01:02:31,440 information about what it's telling you 1687 01:02:31,440 --> 01:02:32,720 you need to be able to see whether or 1688 01:02:32,720 --> 01:02:33,880 not it's knotted so there's a 1689 01:02:33,880 --> 01:02:36,319 topological aspect to it but sometimes 1690 01:02:36,319 --> 01:02:37,880 also you need to be able to touch the 1691 01:02:37,880 --> 01:02:41,960 fiber because your alpaca fiber and your 1692 01:02:41,960 --> 01:02:43,640 something else fiber I don't know I'm 1693 01:02:43,640 --> 01:02:45,520 blanking on South American animals that 1694 01:02:45,520 --> 01:02:47,319 give you fiber your your a pack of fiber 1695 01:02:47,319 --> 01:02:49,480 and your viconia fiber right for example 1696 01:02:49,480 --> 01:02:52,039 might look visually identical but they 1697 01:02:52,039 --> 01:02:53,240 mean different things and you will only 1698 01:02:53,240 --> 01:02:56,440 know if you touch it that's really cool 1699 01:02:56,440 --> 01:02:58,599 um and they were used um as a sort of 1700 01:02:58,599 --> 01:03:00,119 not back Channel isn't quite the right 1701 01:03:00,119 --> 01:03:02,200 word but they were used to encode secret 1702 01:03:02,200 --> 01:03:05,039 Communications during a Civil War I 1703 01:03:05,039 --> 01:03:07,760 think in Lima um but don't quote me on 1704 01:03:07,760 --> 01:03:08,760 that because I haven't got it in my 1705 01:03:08,760 --> 01:03:10,400 speaker notes but that's really cool 1706 01:03:10,400 --> 01:03:13,880 kipu um KH ipu is the search ter that 1707 01:03:13,880 --> 01:03:15,200 you want to look up and they're really 1708 01:03:15,200 --> 01:03:18,760 really cool Q I think it's either right 1709 01:03:18,760 --> 01:03:21,839 yeah yeah um yeah that would be the 1710 01:03:21,839 --> 01:03:23,160 thing that I would want to put in but 1711 01:03:23,160 --> 01:03:24,960 then I was like is it data if you're 1712 01:03:24,960 --> 01:03:26,440 recording like a narrative what 1713 01:03:26,440 --> 01:03:28,599 constitutes data and then I ended up 1714 01:03:28,599 --> 01:03:30,039 writing a whole separate talk about that 1715 01:03:30,039 --> 01:03:32,240 said yes because they used it to 1716 01:03:32,240 --> 01:03:34,559 transfer information regarding in the 1717 01:03:34,559 --> 01:03:37,319 whole empire because was really vast so 1718 01:03:37,319 --> 01:03:39,559 it was just to say like oh you know what 1719 01:03:39,559 --> 01:03:41,279 this province over there they are like 1720 01:03:41,279 --> 01:03:43,319 this amount of person this amount of 1721 01:03:43,319 --> 01:03:45,079 okay so it was very definitely like 1722 01:03:45,079 --> 01:03:47,319 quantifiable stuff we don't know for 1723 01:03:47,319 --> 01:03:50,839 sure how to read it because it was lost 1724 01:03:50,839 --> 01:03:53,319 in history but for examp the things that 1725 01:03:53,319 --> 01:03:55,400 we know is that for example red color 1726 01:03:55,400 --> 01:03:58,279 mhm isans an animal yeah so and depends 1727 01:03:58,279 --> 01:04:01,599 if you just do one not or two notes or 1728 01:04:01,599 --> 01:04:04,079 whatever the heck yeah like they books 1729 01:04:04,079 --> 01:04:06,240 just documenting all the types of notes 1730 01:04:06,240 --> 01:04:08,000 that they just made yeah like I I 1731 01:04:08,000 --> 01:04:09,240 started going down the rabbit hole of 1732 01:04:09,240 --> 01:04:10,720 researching this and then I went down a 1733 01:04:10,720 --> 01:04:12,079 separate Rabbit Hole of like what what 1734 01:04:12,079 --> 01:04:13,440 counts this data what am I going to 1735 01:04:13,440 --> 01:04:14,880 include like there's a reason I didn't 1736 01:04:14,880 --> 01:04:16,640 put cave paintings in here because what 1737 01:04:16,640 --> 01:04:19,760 they encode is not you get into what is 1738 01:04:19,760 --> 01:04:21,200 language and it's like specific 1739 01:04:21,200 --> 01:04:22,520 utterances and you have start having to 1740 01:04:22,520 --> 01:04:24,599 learn about Linguistics and semiotics 1741 01:04:24,599 --> 01:04:26,400 and that's a separate talk which I also 1742 01:04:26,400 --> 01:04:27,680 have but I'm not doing here at this 1743 01:04:27,680 --> 01:04:30,520 conference um it was Inc by the way 1744 01:04:30,520 --> 01:04:32,880 Incan okay yeah um like really like so 1745 01:04:32,880 --> 01:04:35,240 much fascinating stuff um I love all of 1746 01:04:35,240 --> 01:04:37,400 this all of the history of like stuff 1747 01:04:37,400 --> 01:04:40,440 and people and Technology it's 1748 01:04:40,440 --> 01:04:42,240 yeah um all right let me get you 1749 01:04:42,240 --> 01:04:48,630 stickers 1750 01:04:48,640 --> 01:04:51,860 [Applause]