 That's me. That's a photo of me from NDC Sydney last year. I work as a developer advocate at a cloud company called Scaleway and a European cloud provider. We deal a lot with storing data whether or not that's code or media or whatever. I started learning about how and why we do that as a result and going down a lot of rabbit holes. Now we're all here. Before we get started we kind of have to ask what is data storage? What is this concept that I'm talking about today? The best summary I found that I like the most is about in one of the Sherlock Holmes novels, the titular detective doesn't know that the earth revolves around the sun. He defends this sort of glaring gap in his knowledge by saying his brain is like an attic and he has to cram it full of knowledge. In order to know this fact about the earth revolving around the sun, he'd have to throw out something else to make room and he doesn't need to know that in his line of work. So he doesn't keep track of that knowledge. Obviously things aren't quite as simple as that but I like the general analogy. We need to know more things in our day-to-day lives than we can fit in our brains. So we turn them into physical objects so we can keep track of them. This has been true for pretty much all of human history. For example, in the last couple of months, how many of us have had to make a phone call to a number that we did not know by heart? Yeah, quite a lot of us, right? We need to know more things than we can remember and we need to store them somewhere. That's what a phone book does for us or storing things in our contacts in our phones. So this talk is about how we as a species have done that throughout history and the impact in particular that computers have had on solutions to this problem. The other thing I'm going to touch on is about the durability of data storage and that has two parts to it. One of them is about the physical medium involved, like how sturdy are the objects that we create. The other half is about the encodings that we use to turn data into those objects and how and if those encodings survive. So nobody get out your phones. Can anybody tell me without using their phone where this QR code leads? Without specialized hardware and software, this stored data is completely inaccessible. The encoding isn't something that humans can do by themselves in their heads. Even though the object itself is trivially reproducible, this would last forever provided that we could keep copying it. We've seen loss of data in the past due to storage formats becoming inaccessible. We've seen this in examples I'll talk about throughout this talk and there's a concept that folks who are into archival, they're terming it the digital dark age. This is the idea that as our data storage becomes increasingly digital, there's very little left that isn't vulnerable to encodings becoming lost over time as the technology becomes obsolete. So I divided this talk into four sections. We have, going way back into the past, like the Bronze Age, how did people store information before computers? We have got things like obsolete electronic methods, like magnetic drum memory used by IBM and core rope memory knitted by little old ladies that powered the Apollo missions. Really cool, like old historical stuff. We've got current things, things that you'll be familiar with, like hard disk drives, modern magnetic tape storage, flash storage. Lastly, for a bit of fun, I'm going to talk about what the future might hold for data storage and really cool in sci-fi stuff. So, old stuff. Let's look at some things that have survived the test of time. First up we have clay tablets. I promised this in the abstract. I've also got some stickers of clay tablets out at the front at the end of the talk. These were traditionally not fired in a kiln, made of clay, not fired in a kiln, but instead left to air dry. As a result, we don't have a lot of these left today. The physical object is not very durable. Some of the ones we do have, like this one in the picture, it lives in I think the British Museum, this was accidentally fired because a guy's house caught fire. As a result, we know a little about this guy, well, we know a little about this guy. His name was Air Nasir and he was a copper merchant and we know that he wasn't very good at it. This tablet is considered to be one of the oldest known written complaint, one of a few that were found in the ruins of this guy's house, which is hilarious to me, like his house burned down and then like 4,000 years later we're still laughing at him. This tablet is written in Acadian cuneiform. That writing system was lost for a long, long time. Linguists in the 18th and 19th century had to spend a lot of time figuring out how to read this language. So, even though the physical object is very durable in this particular case because his house burned down, the encoding of the data was actually lost for a long, long time. Another method that we've used for a very, very long time and still used today, pen and paper, on the other end of the physical durability spectrum, or in compartment like pigments on papyrus, that kind of thing. Again, the encoding is written language, so we have to be able to understand that language in order to get the data back out of the object. This photo is of the first two pages of the Beowulf manuscript, which was produced sometime between 975 and 1025 Common Era. We don't know when the poem itself was written, and this is the only copy that we have. The manuscript formed part of a library called the Cotton Library, after a guy named Cotton, who had a fondness for old and weird things, bit like me, and for a time this library was stored in this building called Ash Burnham House. Does anyone want to guess what happened to Ash Burnham House? It burned down, it caught fire. Nominative determinism at its best. Unlike with clay tablets, fire is bad for paper, and a lot of the library was either damaged or lost. The Beowulf manuscript suffered a bit of scorching at the time, as continued to deteriorate over time as a result. This is all very interesting. You may be thinking this is a tech conference, this is a tech talk. What does any of this have to do with technology? Now I get to share one of my favourite facts with you. The English word technology comes from the same proto-Indo-European root as the word for textiles. Historically, this makes a lot of sense because textiles were like the site of technological advancement for humans for a long, long time. The things humans need to do in order to survive are growing off food to eat and make enough cloth to wear to use wearers' clothing or use for things like sales for ships. If you look at illuminated manuscripts of medieval art, you'll often see women with drop spindles. That's what that person has, that big, long pole. That's because if you were a woman in this time period and you had five spare minutes, no you didn't. You had five more minutes to be spinning fibre into yarn. Incidentally, this is where we get the English idiom, like spinning a yarn to tell a tale because women would sit together and spin yarn and tell each other stories. That's the original book clubs. So in 2nd century China, people invented the draw loom, which is a machine for weaving thread into fabric and this helped streamline fabric production enormously. The way it works is that you have threads running up and down and then those are the warp threads and then some that go back and forth, which are the weft threads. Each time you make the weaving motion, you need to select some of the upright threads, the warp threads to go forward and the rest to go back and then you pass the weft threads through the gap that's created. I promise this is relevant. They'll make do with boring plain fabric, but what people really like is fancy patterned fabric. You can make that with this kind of draw loom, but it's pretty labour intensive. You have to select those individual threads very carefully from the warp each time based on their colour and it'll be different each time you make that weaving motion. Now, if only we could somehow turn that pattern data and that choice of selecting these particular threads each time into some kind of physical object that the machine itself could use to do that thread selection automatically and that's where we get punch cards. How this works is as follows. When you're picking your threads each time for the weaving motion, there's a little hook that comes down and can pick up the thread or not pick up the thread and the way that it's determined if the hook comes through and picks up a thread is that there'll be a little hole in the punch card. So if there's a hole in the correct place, that thread gets picked and if there's not, it doesn't, ones and zeros. You can automate and programme the process of picking specific and varying threads for each weave and get fancy fabric which much less work and a much lower error rate. These are programmes. Here's some fabric woven with one of these punch card operating looms and the ideas behind this were developed by a few different like artisans and inventors in France in the 1700s ending with a guy called Joseph-Marie Jacquard and he patented a machine that could be added to regular draw looms and convert them and give them this functionality and those are called shackard machines. The patterns from punch cards made in the 1700s are still in use today so that's a big win for data durability, right? Punch cards went on to be instrumental in the history of computing as I'm sure quite a lot of folks already know. We'll talk about that in a moment. The next thing I want to talk about is vinyl audio records. That's because they are the first example of a paradigm that we're going to see again and again in this talk. You have a spinning disc and data recorded on it in a spiral or in concentric rings and then you have a head that you can position on that disc somewhere to read data off of it as the disc spins. Vinyl discs have a single spiral track running outwards from the centre and data is encoded on there by variations on the surface of the disc. The head is just a little needle that runs over that track and turns the texture into an electrical signal and then record players read that signal turn it into noise. Because the data here is encoded on the texture of the disc and the disc itself doesn't have like a protective casing built in. You have to be super careful when handling. Is anyone in this room like a vinyl collector? Yeah, you know where this is going. You have to be really careful with handling them. We get the phrase like a broken record meaning repetitive because if you have a scratch on the disc sometimes what it'll do is it'll cause the head to jump backwards a track and then you end up in this repeating cycle. So physically, as a method of data storage this is physically pretty fragile and the encoding itself can be compromised by common things like dust or grease from fingerprints or scratches. So, now we're entering the wonderful world of electricity and magnetism and computers. We're back to punch cards. For the 1890 US census a guy called Herman Hollerith developed machines for tabulating census data. His data was stored on punch cards. He later founded a company which got merged with a few others and eventually became the corporation we know and love today as IBM. The concept here is the same as with the loom. A gap in the card allows a connection to be made. In the case of the machines that were reading these kind of punch cards the connection was an electrical one. A spring connected to a pool of mercury through that gap to complete a circuit which is obviously very safe to be around a lot of the time, right? Something you might notice is this card has 80 columns on it and this is where we get today's widespread standard of using 80 characters per line if some programs will say you need to have 80 or 79 characters per line so that it will print out nicely in a terminal. Punch cards for data input were output and tabulation were eventually superseded by magnetic storage of various kinds which we'll look at now. This is some drum memory. Now we are cooking with magnets. Magnetic storage works like this. You have some surface of ferromagnetic material that's iron, cobalt, nickel certain rare earth metals and you can align them in a certain crystalline structure and then you can manipulate the magnetic polarity of that surface one way or the other to store data as a binary. Positive for... I don't know exactly which way it is it might depend on your actual system but positive for one, negative for zero for example. Drum memory works by having a drum like a cylinder coated in that magnetic material and a head that can read or write to that material like changing the magnetic polarity of that surface data is stored in tracks which are circles around the outside of the drum. That drum there is I think the Sweden's first electronic computer I believe or electronic storage. Here's another photo of some this is one from a Polish computer called the ZAM41 which is like a great band name I guess maybe that's where ZAM41 got the name Anyway, you can see it's got multiple heads for reading and writing this wasn't always the case you can have any number of heads for reading and writing to this magnetic surface but if you only have one then in order to get to a specific point in the surface it might need to travel quite a long way and but if you've got more then they don't need to travel and that saves you time the trade off there is between manufacturing drums with multiple heads versus the time that you save by not having to make those heads move around the disk as much to access certain bits of the memory which is really important when you're using this kind of memory storage for the computer's working memory so essentially the equivalent of RAM which these were that's what they were primarily used for drums were eventually superseded by hard disk drives for secondary memory storage and primary working memory superseded by magnetic core memory and this is some magnetic core memory this was the dominant form of random access memory for decades and it was so ubiquitous it was often just referred to as core memory or core you've got all these rings of magnetic material that's your iron cobalt nickel and they represent bits and their state is manipulated by these wires that run through them when you run current through those wires they'll magnetise the cores in one direction or another and one thing that's interesting is that even when the current isn't running those cores retain their magnetisation so this is stable non-volatile memory so reading from a core even when your computer is powered off you can look at its working memory but reading data out from this will demagnetise the core so it's destructive readout you can't replicate state from core to core because if you read the core to know what its state is that intrinsically involves demagnetising it one neat fact about this is it was heated not cooled to make it more efficient the constraint here is that you need a constant temperature for it to work best it's a lot easier to heat things up to a consistent temperature than it is to cool them down so if your like home PC ever gets too hot maybe you can like install some magnetic core memory to take advantage of all that extra heat using smaller and smaller cores obviously allowed the core memory to become sort of much more dated dense but the trade off there is then that manufacturing it becomes a very labour intensive process done by hand and for working memory this was eventually superseded by RAM which we use today RAM in your laptop in your desktop PC this is one of my favourite ones in here this is a piece of software core rope memory was a form of read-only memory where the process of writing to it is actually done during its construction like physically threading that wire in and out of those cores is what's doing the programming the topology of those wires represents the software to update that software you've got to do all sorts of like complicated threading and unthreading and this type of memory got the nickname of LOL memory little old lady memory because the process of manufacturing this looked like knitting or sewing or crochet or whatever and lots of different yarn crafts and the most famous usage of this core rope memory was in the Apollo missions Margaret Hamilton who's a very famous figure in the history of computing was at the head of the team responsible for maintaining and updating and developing the software that was encoded into this kind of memory and one thing that made it really useful like what was very useful about it for usage in space travel is that that's very non-volatile like there's no equivalent of like a bit flipping that would be like a thread coming undone or something yeah which is I just think that's really really neat so floppy disks feel really old putting these in the historical section seen here in an eight inch a five and a quarter inch and the three and a half inch sizes these were the latest and greatest in removable storage media from about the mid 1970s to the early 2000s they're called floppy disks because the actual has anyone ever taken apart a floppy disk yeah you know what I'm talking about like the media that they use to store the data is this magnetic sort of like tape like a floppy substance a floppy sort of thing like drum memory and vinyl records they use the spinning disk method this is what the actual disk looks like for those of us who haven't taken apart a floppy disk this is the innards of a three point five inch disk if you pick that up and shook it it would go like and close up it looks like this this is a taken with a I think it's called a magnetic field camera close up visualisation of the data encoded on a floppy disk the bright and dark parts represent bits of the magnetic surface that have been polarised that the ones and zeros and one thing you can see is that they're not completely parallel right they've got a curve to them you know as the track is a curve is a spiral and one thing that this means and this is true of anything that does the spinning disk method I believe is when you address memory by tracks like for the purpose of the example we'll say like concentric rings right you address memory by sector so that's like that pizza slice you can see that and then within that you address it by tracks so I know which sector my data is in and then I go and look at what track it's in here we've got a representation of two tracks within a sector within each sector for each bit of track you're storing the same number of bytes because it takes the same time for the disk to spin across that sector regardless where within it you're looking so you can this is kind of visual representation of what I mean right you've got three bits of you know here's where you've got three bytes stored across this sector and as you spin you'll be checking at the same interval so on the outside edge the data is physically much more spaced out and this is inefficient you're making inefficient use of the magnetic surface Apple actually experimented with fixing this in early Macintosh computers by varying the rate at which that disk span when it was reading from the edge versus the centre this required unique drivers unique circuitry and unique hardware disks written with this method weren't readable by drives that didn't use this method and it just became economically unfeasible to maintain that as a method of writing to floppy disks eventually they reverted back to the constant angular velocity model that gives you this inefficiency at the outer edge that is something that happens with I think hard drives there are some people who've done the same with hard drives like varying the angular velocity based on where you are physically on the between the centre and the edge of the disk and it just never ends up working it never ends up being worth it which is just an interesting pattern that you see again and again with spinning disks I think some vinyl might do that I'm not sure so before computing environments were regularly networked floppy disks were like the go to method for transferring data between one computer and another this legacy lives on with the floppy disks having recone the global icon for the save button when I was working on this talk and I was giving out these floppy disk stickers I think about 15 people over the course of two days went oh you got stickers of the save icon and they were all like guys in their 50s making the sort of dad joke about it so the floppy disk the legacy lived on in the save button but all good things come to an end eventually other data storage formats became more competitive in terms of cost of manufacture and density of storage so by 2007 only 2% of computers sold in stores had built in floppy drives however this article is from October 2019 if you ever worry that the technology that you're very familiar with and that you know best is getting old maybe you need to reskill in AI or WASM or whatever don't sweat it somewhere in the world probably the US government will still be using obsolete technology for years and years and years they're not quite dead yet as well I went on Amazon I had a gander at some of the external floppy drive disks you know plug into your computer via USB I couldn't find one with a USB-C connector but if I wanted to read a floppy disk onto my MacBook I could Wikipedia calls floppy disks obsolescent which I think is just an incredible incredible world so we are now into current stuff I'm going to take a very brief water break here as a side note here do we have anyone who has never used a floppy disk in this room they're great you should get one and own one just to have it cool so current stuff that I'm going to assume that everyone in this room has used at some point in their life optical disks optical media more spinning disks so does anyone here have a PC with an optical drive in it a couple does anyone have a games console that takes CDs actually about the same you get a couple hands for the first then everyone's like oh yeah I've got a PS4 and it takes CDs I debated putting them in the previous section because most PCs now don't come with optical drives as a default I don't have that on my Mac but they are still widely used in consoles so I'm grouping a bunch of things together here under the heading of optical media laser disk CD ROMs, writable CDs all of that stuff they all use the same paradigm as vinyl and floppy disks so I'm going to turn around and something reads data out of them on optical disks data is encoded with physical bumps and dips on a polycarbonate layer within the disk and then that means that if you want to mass produce disks you can actually have a stamp and you go and physically stamp the disks and produce that texture this means the disks aren't then writable but for example if you've got CDs of music the month on CD I bought 18 or whatever that was how that was produced then that layer is coated with aluminium not always aluminium but a reflective layer to make it readable by a laser and then that's how the computer reads it out it's got a laser diode that can read the bumps and dips on the surface by how reflective they are so here we can see the detector which is the lens from an ASA laptop optical disk drive I believe specifically this is a CD drive CD ROMS Read Only Memory were developed by Sony in 1984 and it wasn't until 1997 that we got rewritable CDs this is a rewritable CD and you can see that its iridescent layer is like a lot darker than the one that we saw a couple moments ago and that's because the metal that's coating that textured surface is not aluminium it's made of a silver, indium, antimony, tellurium alloy I'd say in that three times fast and what that does is it allows the laser that can read from the disk to also melt that metal and change its reflectivity thereby altering what will then be read out of the laser subsequently so it can use the same laser to read and write to the disk but the way that then once you have done that to the disk it's very hard to undo that you have to heat the whole disk up and remelt all of the metal at once and return it to its crystalline structure and so that's why if you're going to burn a CD that's why it's called burning because it's actually physical heat was involved in rewriting disks like that Reading data out of an optical disk requires its surface to be legible to the thing that's reading it much like with vinyl they're very susceptible to things like scratches fingerprints you can also straight up break them I just included this because I love it it's a video, it's a clip from it's a clip, it's a screenshot from the slow mo guys spinning a CD very very fast until it breaks and then taking filming that with a very high speed many many frames a second camera 170,000 frames per second which was like state of the art back in 2010 or whenever they made this video that's CDs now we want to magnetic tapes here we have an example of a data storage medium which is not random access so with all the other types we've looked at you can sort of, if you want to read a particular bit of data you can go and do that in roughly linear time with magnetic tapes you can't because you have to scroll through to get to the bit that you want to look at like you have to rewind VHS and cassette tapes in order to get to a specific bit of them this is true of data stored on any kind of magnetic tape so the tapes are made of ferro magnetic material and we all know by now how data is encoded onto those poloise that magnetic material the hardware that reads and writes to this tape is called a tape drive the first magnetic tapes were used to record computer data back in 1951 for the Univac one and IBM quickly followed suit set the standards for the industry magnetic tapes have been in pretty constant use ever since despite that drawback of having potentially long access times they're very cheap, they're very durable magnetism wise if you're not reading a writing to them that magnetism is very stable and they also you can physically remove them from the computer in a way that you don't with like most hard drives these days you can physically air gap your data from a system that can read and write to it so if you've got data in a data centre that's just sitting in a room somewhere that's safe from cyber attacks I guess however you might need to wait a while before you can actually access the data that's on it like if you've got so a lot of cloud providers and this is the link with scale way essentially is that a lot of cloud providers use these kind of magnetic tapes to store data, cold storage of data and sometimes there are conditions on that where you need to wait a little time for I've requested some of my data from the cold storage facility and there's a wait time because somebody needs to go and get the physical tape put it into a computer, read that data out and transmit it to me, it's not as instantaneous as it might be or as it might feel from other kinds of data storage but that's an acceptable constraint for cold storage because you don't usually need cold storage data that quickly, that's actually an acceptable way to store your data one last note on these one of the most famous examples of the digital dark age is from the 1975, so in 1976 Viking NASA Viking lander on Mars, he's got Carl Sagan next to a replica, he didn't actually go to Mars and this lander recorded a bunch of data onto magnetic tapes NASA got the tapes back and then left them in the cupboard for 10 years when they got round to trying to read them, all of the original engineers original programmers had either left NASA or died and they could not read that data off it took ages, it took like months and months to reconstruct the data format that had been used to record to these tapes and involved looking at the physical recording mechanism on the lander that they still had and if people had written down the data format that they used to encode their data that wouldn't have been necessary so durability of encodings does matter more spinning discs hard drives, we all know them use them, love them you can get them built into most personal computers unless you're like an SSD only person and you get external ones that can fill with lots of pirated media I looked on Amazon the other day and it looks like 20 terabytes is about as high as it goes commercially right now I'll hold a lot of movies though Wikipedia claims that 26 terabyte drives are available can't find them so these work using the same paradigm as floppy disks and optical media things spins round, another thing wins data off it the medium is again magnetic storage so don't put your electromagnets too near these once again it's an IBM thing that they pioneered the first two came out in 1957 storing a whopping 3.7 megabytes of data in the photo above you can see two of them they had 52 disks each and one arm holding two read-write heads IBM kept iterating and improving and they briefly flirted with a model that had one read-write head per track the IBM 2305 meaning that they didn't lose any time to heads having to move around the disk to find data in specific parts these turned out to be incredibly expensive and were discontinued over the 1980s hard drives went from being expensive add-on to a PC to being standard issue now modern hard drives have got to the stage where the area of the disks that's being magnetised is actually so so so small that they risk losing that magnetic state due to an effect known as superparamagnetism whereby heat can cause magnetic magnetised material to flip its polarity in areas smaller than 50 nanometres so for reference the thickness of a piece of like copy paper is about anywhere between 70,000 and 180,000 nanometres so we're talking absolutely tiny tiny scale here to counter this disks now actually have two layers of magnetic material separated by a three atom thick layer of ruthenium which is not magnetic and those two magnetic layers have opposite polarities and they reinforce each other I just think it's really cool that we've sort of bumped up against what seems to be a fundamental limit of the universe of how small you can make magnets in the purpose in the service of storing data I just think that's really really cool so this is a data bunker 25 metres underneath Paris owned and operated by Scaleway I don't know where it is in relation to the catacombs under Paris I don't know how close it is this is where we keep our cold data storage and we don't use magnetic tapes for that we actually use hard drives and that's unusual because magnetic tapes are better suited in a lot of ways but with hard drives one of the things you can do is you can alter the way that you actually write data to them there are two different methods one is called parallel magnetic recording which is what most hard drives use and the second one is called shingled magnetic recording and the trade off is that you get a much much higher data density but it's much much more difficult to write to that disk in a random in a sort of random access way you have to write data all at once and then you are able to read that data much more efficiently and store more of it on the same disk I just think it's neat that we've got this weird method of writing to hard disks that we use at my company and I had to go on Wikipedia and learn about it in order to know what's even going on right, last but not least flash memory you have some of this in your pocket probably it's in smartphones, tablets if you've got a UB key to a FA token that's got some flash in it too but if you try using a circuit component called a floating gate MOSFET which is extremely complicated is essentially able to store electrical charge that's ones and zeros flash drives have a few limitations for a given block of storage you can only really write to it all at once not in patches like the hard drives in scaleways called data storage they also have a lifetime limit on how often they can be written to because those MOSFETs wear over time and they'll leak charge over time because they leak charge over time a problem where if you write to a flash drive then leave it in a draw for 10 years like NASA did with the Viking lander tapes it will leak that charge over time and the memory will become corrupted yeah, so they're not durable over long periods of time due to the physical nature of what you're doing to store the data right, and SSDs are flash storage as well the read-write limitations over time are gradually improving so don't worry that you're constantly writing to your SSD and it's going to wear out that's not really a concern it's just interesting right, future stuff because I have about 8 minutes or something left we've learned all about the history and current state of data storage we know a bit more about how and why the various types use various purposes all of the current methods we talked about they're probably going to get incrementally better over time like we'll improve that lifetime wear limit of flash memory hard drives will keep getting bigger and better but none of that is like cool and sexy I want to talk about things that are futuristic so DNA it's very very data dense microscopically small cells can hold the entire genetic code for an entire species a team at Los Alamos National Laboratory created an encoding the adaptive DNA storage codex for turning computer friendly binary into the DNA friendly pairs of A, C, T and G bases there are a few other encodings out there you can go read about them with various trade-offs the idea is you create synthetic DNA representing your data and that can be stored very durably at low temperatures for decades we can do this today, there is DNA data on the moon courtesy of the bearish heat lander in April 2019, like the lander did crash so maybe that data is actually lost but if it isn't then the foundation that built and transmitted that data sent the lander estimate that that data could still be stable in billions of years that's incredible so DNA I was wondering if it was people were storing data in DNA because of its ability to replicate with very low error rates couldn't find any information about that but I think that's a really cool idea and I'd like to see people explore it 5D optical data storage this is also known as Superman storage the 5D is a marketing term the idea of the five dimensions is that you've got the three dimensions and then what rotation it's at and how close you are to it and those are your five degrees of freedom this is a bit of transparent non-photo sensitive material like a chunk of fused quartz or glass and you take a laser and engrave it this is a marketing play but this photo above is of the universal declaration of human rights engraved into a piece of glass in the UN symbol because you can also make them look pretty both Hitachi and Microsoft have worked on this concept the current bleeding edge of it is in the Southampton University in the UK the data isn't going anywhere it's very durable physically unless you drop the piece of glass which can happen the encoding used to turn the data into engraving isn't yet settled as far as I can tell maybe the declaration of human rights here is like just there in like seven bit ASCII if it is and we still know how to decode that in a thousand years we will be able to use this piece of glass to retrieve it that's super cool right very last thing just to get really really weird with it do we have any condensed matter physicists in the room? cool good nobody here to call me out when I get things wrong this up here isn't a time crystal there are no good photos of them unfortunately it's a space crystal and like flash memory you have this in your pocket probably that's silicon and it's the thing that makes it a space crystal is that it has a recurring structure in space this physical structure repeats over and over again time crystals are the same but instead of space they repeat in time there aren't actual photos of them but we have made them in labs this gif kind of gets the idea across a time crystal is an object whose lowest energy state its resting state is one of motion this doesn't break the conservation of energy because there's something something quantum physics they were theorised to exist in 2017 2012 by a theoretical physicist called Frank Wilczak and in 2021 a team of scientists at the University of Hamburg actually made one in a lab it didn't look like what you'd think of as a crystal it was a Bose-Einstein condensate which is a state of matter you can get when you super cool boson gases I hate when I open my freezer when I've actually made a Bose-Einstein condensate again the reason that they are in this talk is because clever people think that we might be able to use them as some sort of memory storage for quantum computers that's currently an open problem in quantum computing there's in quantum theory there's a thing called the cloning theorem which states that you can't duplicate quantum state you can only transfer it and that means that any read operation from something being used to store quantum state would be destructive just like core memory where when you read from it you do magnetised it and lost the data yeah which I just think is really cool like you see these sort of same things coming up back again but yeah I read the Wikipedia article for time crystals because they're under the Wikipedia list of data storage formats or not formats but options and I had a brief moment where I comprehended it had an out of body experience and then I completely lost it again but they're very very cool and if you want to have a real trip go read the Wikipedia page about them so conclusions we've learned about storing data some of the complaints in the bronze age all the way to encoding data in our DNA and becoming human information cyborgs it's been an incomplete and whistle stop tour most of what I wanted to impart with this is that humans are really cool and creative and we're very good at making increasingly complicated hardware but if there's another takeaway from this it's that data is useless if you can't read it when we make decisions about how and why we store data it has to be with durability in mind if that team with the Viking lander had written down their data format and their encoding they wouldn't have had to spend months and months with lots of different engineers new engineers learning how to read it again you have to bear in mind all of this stuff is useless unless another human which might be you in the future can figure out how to use it that wraps me up that QR code is not a Rick roll it is a link to a little form where you can give me feedback if you would like to and we have two, three minutes for questions yes thank you very much for the talk hi, thanks for the great talk are you hopeful about the current projects to preserve our data such as the internet archive do you think they will be successful at saving current data in thousands years so the question is I don't know how audible that was to everyone else am I hopeful about projects to preserve data for let's say the next thousand years like digital preservation I think there's a lot of really interesting work going on in that space so the Superman storage thing they are designing that specifically with durability of data in mind that's an explicit goal and I really like that that is something that research teams are focusing on I think one of the problems is we just generate so much data these days and a lot of it is useless but there was an article that came out recently saying that only 13% of video games are available to play right now or you can even play them and that has been lost which is really sad and I think there's I think a lot of the efforts to preserve data are like quite grass roots and not organized and as such they're always going to be kind of playing catch up against the constant churn of data that we're producing yeah, yes and no I think there's lots of really interesting work going on in that space and I think that's going to yield a lot of useful things I also think that those efforts are probably underfunded and they're not rewarded by our current economic system so they're going to be at a disadvantage Does that answer your question? Any more questions? Okay then let's thank Eli for his great talk Thank you