26

I remember an episode in Star Trek TNG (probably first season) where Data is describing some of his internal workings. Among them, he lists that he is equipped with "12 quadrillion" bytes for memory storage. If my math is right that is 12 petabytes.

It seems to me that is severely short for the amount of data he collects every single day.

I found this related question:

Can and Does Commander Data make frequent backup's of his programs and memories?

However, it seems that Data likely did not back up anything that he learned. I do remember him stating a few times that he forgets nothing.

So this is the question: Is 12 PB enough storage for an android like Data (assuming none of it is taken up by OS and programs)?


Consider some of the implications about massive amounts of data on this article.


Well, I just found this which says it is 800 quadrillion bits! That is quite a bit more (pun intended) but I am certain that the episode I am thinking of said 12.

  • 12
    Maybe that was just the L1 cache? >_> – jono Jul 9 '13 at 20:19
  • 2
    @fredsbend For a scale, the NX-01's memory was measured in exabytes during the Xindi arc. – Izkata Jul 9 '13 at 20:35
  • 4
    The level of compression is very important in this case. If for example Soong learned some sophisticated compression techniques(perhaps aided by the fact that seemingly P=NP in the Star Trek universe) then it is highly plausible that that much memory is enough. – NominSim Jul 10 '13 at 2:39
  • 1
    @fredsbend To be honest, if Data had an L1 cache in the petabyte range I don't think he'd need RAM. Or rather, it's hard to imagine his design could allow that much data in the L1, and still stand to benefit from further levels of caching. – jono Jul 10 '13 at 23:46
  • 5
    He shouldn't have needed any more than 640k. – Stick Feb 27 '14 at 23:55
17

The most definitive answer came from the episode "The Measure of Man". In his trial to determine whether he is Starfleet property or an actual individual, Data states his computational capacity as possessing a 800 Quadrillion bit storage memory, with computational speeds of 60 trillion operations per second.

However, throughout the series, his positronic net is under constant upgrades and tinkerings from every character from Geordi La Forge in multiple episodes to the Borg Queen in "First Contact". So, this measurement of Data's "brain" should be considered cannon for only this particular episode (in my own opinion).

However, regarding your question whether this is enough "operating space", I would argue that, since Data is constantly devouring new information, often for the sake of "humanizing" himself and the fact that he isn't ever seen having any difficulties in his own operating capacity, that he more then like has enough memory to handle all his functions. Soong built data to last, and if need be, Data would most likely have options to either "Dump" parts of his memory into storage units, or to increase his memory capacity through the use of physical upgrades (Much like his emotions chip). It is also important to note that Data has been fully functional for nearly 60 years before becoming part of the Enterprise crew and that he he "died" later through his own heroism and not through a lack of memory space. It is probably safe to assume that if there going to be any memory problems, they would have sparked up somewhere between these two periods.

An interesting note is that the Doctor (Voyager), as a hologram had significant memory storage issues in his holomatrix and at least once was forced to pretty much be reformatted from the ground up in order to save his program. So there have been times within the series cannon that this "dilemma" was a serious one. However, since Data is not based on Starfleet technology, and is an Android with a physical matrix, the same limitations could not be applied.

There was always a problem in the series regarding the use of Byte statistics to clarify processing power. If anyone remembers the Voyager three-parter "Futures End", Starling downloaded a WHOPPING 3,000 GIGABYTES of information which also happened to include the entirety of the Doctors program. It is laughable by today's standards of course but I believe they fixed this later in both Deep Space Nine and Voyager by simply changing the terminology to "Quads" in order to address the issue of cross comparing to modern systems.

So if we are unwilling to take the above solution to Datas storage issues as a possibility, the meta solution to this conundrum would simply be to switch the terminology, which is most definitely what Star Trek Writers/Directors would do today if the question was posited to Data today.

See the below video where he answers the question regarding his specifications-

  • I stated that Data was not based on Starfleet technology. That doesn't not mean that he was not based on HUMAN technology. I stated that because his creator, Dr. Soong was a civilian member living on a small colony. There are multiple cases of issues popping up with Data syncing up to existing Starfleet technology and, being far superior in construction, Starfleet attempts to understand this technology through analysis of Data. – Robo Stalin Jul 9 '13 at 22:53
  • 1
    "800 quadrillion bits" would be about 100 petabytes. I think that would be enough to last quite a long while. Originally, I thought he had said "12 quadrillion bytes". The memory dump is likely. Also, we shouldn't think that Soong would build everything about Data so magnificently but then fail to give it enough memory to function over its lifespan. +1 – fredsbend Jul 9 '13 at 23:03
  • quads and kiloquads were a TNG invention - it's a bit unfortunate that the writers didn't start using this kind of obscuration from the outset, but I guess they wanted to avoid calling a rabbit a smerp. – HorusKol Jul 9 '13 at 23:40
  • 2
    To be fair, 3 terabytes would have been a crap load of data at the time of the episode :) – Jared Jul 10 '13 at 1:28
  • 3
    Wow, just typed "800 Quadrillion bits" into Wolfram Alpha and for comparison it listed "purported storage capacity of the character Data in Star Trek: The Next Generation ( 8×10^17 b )" – Dracs Jul 10 '13 at 3:50
5

This question depends heavily on what it is Data is expected to remember, and how long he is going to live. Naturally, if he lives forever then no amount of storage is enough. Additionally, if he records an exabyte a minute he's going to run out of storage almost immediately. But we can make some assumptions and see how they turn out.

First, we're going with the limit that Data can store a total of 800 quadrillion bits as mentioned in The Measure of a Man. This is exactly 100 petabytes, or roughly 88.8 pebibytes. For comparison, this is about 250 times more memory than that taken up by every book ever written in any language, or about 2,000,000 blu ray discs. We're assuming that this is memory intended exclusively to record the day to day happenings of Data and won't need to be shared with other things relating to his programming and function.

Now the real question is: How much data does Data save every day? Things like smell, taste and touch are hard to quantify, but sight and sound are not. If we assume that Data simply keeps a running record of everything he sees and hears, then the question instead becomes how high his resolution is, how many frames per second, and what codecs he use. For simplicity's sake we'll assume he saves video at a 16-bit RGB standard (likely much higher), giving 6 bytes per pixel.

Uncompressed, a 1280x720@25 video stream would require ~138MB per second, or ~12TB per day. Assuming an audio bit rate of 320 kbps would only add some ~28GB per day, so it's pretty negligible, giving Data the capacity to store about ~23 years of his life.

This is obviously not the case since not only did Data live for at least 40 years, it would also be a humongous waste to not compress anything. Modern lossless video codecs could bring those size estimates down by as much as 50% (though the average compression would be much lower than that), bringing the daily storage requirements down to ~6TB. Lossy video codecs could bring the estimates down to a minuscule fraction in the gigabyte range, potentially giving Data thousands of years of storage.

1280x720@25 with 16 bit colors and 320 kbps audio is pretty low quality given what we've seen of Data's capability. But there are things we can speculate that his codecs are capable of, which would allow us to up the resolution and frame rate significantly, and still give Data a reasonable life span. For example, the codec could be capable of removing the undoubtedly massive amounts of redundant frames, compressing the information in those that are repeated hourly or daily or weekly or monthly but not in sequence. Text could be transcribed, repeated pictures and sounds could be stored and referenced. If an oft seen room was mapped as a 3D model, frames taken in those rooms could instead just store Data's position and viewing angle, then remove everything in the frame but the deviations and additions, like people and furniture. There's a lot of things you can do with 24 hour video, if it's recorded from the perspective of someone with fairly undeviating routines. Some of it is beyond our current day capability, but surely not beyond someone who can make a hard AI.

There's a lot of other things Data probably saves as well: His thoughts, other sensory readings, conclusions drawn, but if they can be transcribed as text their additional storage requirements are negligible.

TL;DR Given a few (a lot) assumptions about what kind of data Data stores, and the compression capabilities of the 24th century, 100 petabytes should be more than enough to keep Data going for a long time.

  • +1 for mentioning compression schemes. That was the first thing that I thought of as well: it may not be how much he can store but how efficiently he can compress and decompress the information. Given that this is the 24th century his technologies are most likely far more advanced than anything we can logically explain. Plus, his processor may find it more efficient to use high-compression and decompress as needed -- thus keeping storage use to a minimum -- then to store everything as lossless. – redknightalex Jul 10 '13 at 22:58
  • 2
    +1 for mentioning 3D modelling. There's no need to record everything he views in compressed audio/video if he has a memorized 3D model of the Enterprise, contents and crew and can recreate his P.O.V. and events that I presume he has memorized with fairly high fidelity. – ghoppe Jul 10 '13 at 23:13
2

At some time on April 25, 2012 the Library of Congress had over 3 petabytes of data digitally stored. Assuming you use data compression methods similar to what is available today, using loss-less codecs, I would say that 12 petabytes would not be anywhere near enough space to perfectly record every moment of every day for an android with no known preset limit on the length of it's life.

  • Depends on how the data was stored. If it was stored as video, no, probably not. If it was stored as audio, then maybe. If it was stored as text transcripts, then most likely it would – The Fallen Jul 10 '13 at 13:18
  • I'm not sure why we must assume that Data perfectly records in full HD audio/video every moment of every day of his life. Has that been stated in canon? My assumption was that he parses and analyzes during some downtime. Perhaps he has a full archive of the last day or even week, but only stores important events in full video. The rest is a semantic summary (which may well be the equivalent of a novel a day, but easily compressed and stored.) – ghoppe Jul 10 '13 at 22:25
  • @ghoppe It is going to bug me now until I can find the episode, but there is a moment when Data and Geordi are comparing how human memory works compared to Data's memory, and I want to say that Data says he remembers everything and is puzzled that humans cannot. – Xantec Jul 10 '13 at 23:29
  • @Xantec I believe you, but I would like to point out that "remembering everything" is not the same as "recording everything". – ghoppe Jul 11 '13 at 16:10
  • Indeed, it's never suggested that Data has an instant replay, he may simply detail the important things in a database - a snapshot of the person talking along with their 3D coordinates, stance, movement and speech (noted with inflection, timing etc) can take up significantly less space than a video of the event, and that could be recreated later. Many computer games use this technique even now to record replays... they note what happened and then re-create the video on demand from the events. – Jon Story Dec 2 '14 at 12:28
2

As another hint whether Data's storage capacity is sufficient or not, I quote from Scientific American that the human brain is estimated to have a storage capacity of maybe 2.5 petabytes.

According to Memory-Alpha, Data's brain simulates a human one in that it works as a neural network. While our understanding as to how exactly memories saturate a neural network is rather limited the quotes in the other answers seems to suggest that Data has more than sufficient memory to keep a life in memory that may even span many centuries.

  • Data's brain simulates a human one? I can believe that its structure is based off human brain structure, but not simulate it. He remembers, or rather records everything. – fredsbend Jul 10 '13 at 16:18
  • Yes - I wouldn't imply it's the only thing it does. – flq Jul 10 '13 at 18:03
  • 1
    @fredsbend "He remembers, or rather records everything." — citation, please. It seems absurd to me to build an Android that's simply a video recorder. It makes more sense that he analyzes what's important to remember, and stores the information semantically and not in full video. Like a video game, he may even be able to have all the physical characteristics of the Enterprise and crew committed to memory and reconstruct a past memory, rather than store the position of every photon that hits his eyes. – ghoppe Jul 10 '13 at 22:33
  • @ghoppe in the very first episode he not only quotes Picard exactly, he speaks it exactly as Picard did and in Picard's voice. You can split hairs if you want, but that is recording, not memory. – fredsbend Jul 10 '13 at 22:50
  • @fredsbend I presume data has memorized the voice patterns of everyone he deems important. I'm certain he can recreate anything he wants in Picard's voice. Indeed, I remember the episode where he hijacks the Enterprise to visit Dr. Soong and speaks commands to the computer that Picard clearly wouldn't. Even if as you maintain, that particular instance is a recording, I still find it implausible that Data records and saves every second of his waking life. – ghoppe Jul 10 '13 at 23:03
1

I arrived here after a search for "how many times has Data saved the Enterprise?" One thing that I believe no one has mentioned is that you are all basing these assumptions off of Binary Computing, 1's and 0's. Also, the architecture you are trying to scale to fit to Data is all x86 Based.

I know for certain TNG era computer systems are no longer Binary and are definitely not x86 based. I can recall from at least one movie or episode, I believe it was a Voyager time travel episode with the doctor visiting Earth or a TNG episode that they had to switch to an "archaic" binary format to interface.

Surely that far in the future, even Quantum computing is a thing of the past. Storage capabilities and functions likely don't even resemble what they do today. A modern desktop these days could probably store every piece of digital data in existence from Circa 1970 and that's only 40 or so years and still all Binary.

  • 1
    They do have a quantum computing measure (the quad) that they use regularly in trek, however Data gives his own computational capacity in bits, not quads. "800 Quadrillion bit storage memory", etc – Valorum Dec 25 '14 at 8:34
  • I recall a line in this episode where the cloaked aliens aboard Voyager call the computer "A simple binary system." I'd have to watch it again to be sure. – fredsbend Dec 26 '14 at 20:11
  • @Richard I thought the introduction of "quads" into the trek universe was to make the content more timeless. "quads" aren't anything real, so we can use our imaginations to suppose that it is something better than what we have now. AFAIK, quads are never defined in the trek universe. – fredsbend Dec 26 '14 at 20:26
  • @fredsbend - Yes, and also to imply that in the future they use measurements that we've not even heard of... – Valorum Dec 26 '14 at 21:02
  • 2
    @Richard We're not talking about programming here. We're talking about the fundamental structure of the storage devices which dictates the way machine code works, which is binary for our devices. – fredsbend Dec 26 '14 at 21:52
1

Positronic brains are hypothetical synthetic simulations of real brains, which are not really binary systems.

So the 100 Yottabyte doesn't even make sense, unless Data's brain is hybrid binary computer and a positronic brain, that is both CPU and storage. This would mean that the storage capacity is merely a database system to quickly access and store information, but that he additionally also has a conscious memory that is potentially much bigger but takes a lot of time to develop because of it's complexity.

  • Binary is currently used in computers in order to assure a high degree of accuracy in signalling a device's current state. If devices could be created to be accurate to a certain degree, the output could be whatever number of states that could be efficiently certified. At side note: most binary devices are capable of three states: high, low, and high impedance. The third state is so that a number of devices can be connected to the same address or data line. – Howard Miller Jan 26 '16 at 23:43
0

Data does say in that episode that he has 800 Quadrillion bit storage memory... This could mean 800 petabits, or it could mean 800 yottabits, depending on if he was using the long or the short system. So he either has 100 petabytes or 100 yottabytes. I would say the latter would probably be more likely, seeing as it's 109 times bigger. It would seem that the 24th uses the long system probably because it makes more sense and the non-English speaking people use it. I also agree with the upgrade theory of Data's memory.

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.