PRISM and Bill Gates’ ‘documented life’


The future changes so quickly that Bill Gates had to write the same book twice. When The Road Ahead was first published in November 1995, he was still tepid about the internet, not quite prepared to make it synonymous with the ‘information highway’ that he hyped throughout the book. He had, in fact, changed his mind at least six months earlier, when he released the company memo in which he charted a new course for Microsoft centred around networking and the internet. But it was too late to change the book. He finally did so ahead of the paperback edition of 1996, and so for a while the future as described by the world’s most important technologist was available in two quite different versions.
The Road Ahead, in either edition, has little more than curiosity value now. But in it Gates made a throwaway prediction that still resonates. He described how, armed with one’s ‘wallet PC’ (what we’d now call a smartphone), people might choose to lead a ‘documented life’: that is to say keep a constantly updated record not only of their interactions with the internet but also of all of their everyday conversations, as well as of data such as body temperature, blood pressure or atmospheric pressure. While admitting to finding the prospect a ‘little chilling’, Gates speculated that this running database might constitute an ‘alibi machine’ with which to guard against false accusations. ‘On the other hand,’ he concluded, ‘if you were guilty of something, there would be a record of it.’

datacentre

The record is out there, we know that now. Not every conversation, but most conversations. The what and the where and the how. So much personal information that it uses the last available name for a unit of data: a yottabyte, or one trillion terabytes. Nobody has even bothered to name the next one yet. The picture of the facility that the US National Security Agency is building in Utah, above, was attributed by Wired with delicious irony to ‘Name Withheld’. Who knows if the fortified buildings will show up on Google Maps, so that we wired citizens of the world can pinpoint the exact location of the last and most secure backup of our Skype conversations, email address book or family photos?

Apple, Yahoo!, Microsoft, Facebook, Google: if its leaked internal documents are to be believed, the NSA is able to access all of their servers directly and at will as part of its PRISM programme. This wouldn’t be surprising, or at least it shouldn’t be, for anyone who is familiar with the word Echelon or the phrase Room 614A. It’s an idea with documented precedents in the recent past, and since then the appetite for total surveillance of the nations involved – the US above all – has hardly diminished. What remains to be determined is the precise nature of the system, its legal boundaries (if it has any), its technical limitations (if it has any). These are far from unimportant questions, but it’s easy to get lost in the details. Yet global surveillance is nothing if not a big picture. So much information that the quantities cannot be imagined, information containing not just digitally typed words that a search engine could make ready sense of, but numbers with innumerable different meanings (one person’s phone number is another person’s bank account balance), pictures, audiovisual recordings. Everyone’s documented life. ‘Spying’ doesn’t begin to describe this. I agree with Robin James that we need a new theory of listening, as well as of watching, or reading.

Consider again Gates’ ‘documented life’. What kind of record is it, and how would you access it? To replay it would take a lifetime, by definition. There is no memory, scrapbook or diary that contains everything, the very idea of biography is based on selection. But we have the technology now, an architecture centred around consumer electronics and social media for collecting all personal data, from the objective to the psychological. I was in such a place, at such a time, and my GPS signature so confirms it. My mood at the time , as recorded on Facebook via the drop-down menu provided, was ‘happy’.

Even if you could replay a documented life, the data, in isolation, would have little or no meaning, even and perhaps most especially to you. It makes sense only in aggregate, for corporations and the surveillance state, and the kind of sense it makes is incommensurate with individual human goals and worldviews. Your being happy in a particular situation, for instance, might be of interest to marketers, so long as they can find enough matching emotional profiles to enable them to construct a workable model of happiness – what it looks like, how best to cater for it. Law enforcement and intelligence agencies will look at different patterns, or maybe at the same patterns in different ways. They will seek not comforting repetition but dissonant anomaly. Something that suggests your record ought to be pulled out and analysed. Give a single month of someone’s perfectly documented life to an organisation that has the right software and expertise, as well as the man hours to devote to the task, and they’ll make sense of it in no time.

boundlessinformant

As several commentators have noted, when Barack Obama said on Friday that ‘nobody is listening to your phone calls,’ he likely wasn’t lying. For one thing, nobody cares about the content of your conversations. Not unless you are already the target of an investigation. But they care very much who you re talking to, where ‘who’ is narrowly defined as a set of coordinates – a phone number, a location – that can be plotted on a map. Boundless Informant, one of the secret programmes whose existence was revealed last week by the Guardian, is all about plotting the billions of pieces of intelligence gathered by the NSA on a world map, with different colour gradations indicating the relative intensity of the information collection. To what precise set of ends, it’s difficult to say, but the immediate effect for the layperson is to highlight the total reach of global surveillance, its boundlessness (although its not without internal boundaries).

The map suggests that intelligence is still being gathered with a degree of selection, but bear in mind that the Utah data centre is not finished yet. When it comes on line, in September of this year, it might be possible to fulfil the vision suggested by PRISM: a world in which all social data that circulates is captured. It is the ultimate banality of our time to suggest that we engage in self-surveillance of our own volition, as a freely made consumer choice. Gates said of the electronically documented life that it would be ‘the ultimate diary and autobiography, if you want one’ but in fact soon enough being on Facebook or on an equivalent service will be about as optional as it was thirty years ago to have a phone (if it isn’t already). Yet how else you describe those geolocated tweets and status updates, or those situating statements – I’m at the supermarket, I’m on the bus – except as a participatory compulsion, a desire, a need perhaps to be plotted on such a map? That it should be a statement of banality is all the more unsettling. We are, each of us, most of us, boundless informants.

Giovanni Tiso

Giovanni Tiso is an Italian writer and translator based in Aotearoa/New Zealand and the editor of Overland’s online magazine. He tweets as @gtiso.

More by Giovanni Tiso ›

Overland is a not-for-profit magazine with a proud history of supporting writers, and publishing ideas and voices often excluded from other places.

If you like this piece, or support Overland’s work in general, please subscribe or donate.


Related articles & Essays


Contribute to the conversation

  1. As several commentators have noted, when Barack Obama said on Friday that ‘nobody is listening to your phone calls,’ he likely wasn’t lying. For one thing, nobody cares about the content of your conversations.

    You are mistaken here. The metadata is almost useless unless they know what is being said. Much more valuable information in the actual content of the calls.

  2. I worked for a while for a small startup company. One of its technologies was a “video feature segmentation engine”: a piece of software that used machine vision algorithms to extract features of interest from a drone camera flyover video – essentially a top-down view of some territory – and “segment” them from a largely static and strategically uninteresting background.

    The first idea of the technology was to recognise what was interesting. The second was to massively compress the amount of storage required (the flyover video, sized in the tens of GB, could be converted on board the drone to a largish static photograph with a few moving features on it) so that it could be efficiently wired to a base station for further processing.

    Even then, however, it was anticipated that this would be embedded in a human-operated application that allowed a military analyst to flag extracted features for further investigation (the company actually sold such an application which also had some other features).

    While the mere fact of capturing data is a great concern, and when it comes to human verbal communication, the algorithms required to extract “features of interest” are simpler, there is probably presently a huge backlog of difficult R&D problems to be solved before PRISM represents the “worst nightmare” surveillance of science fiction.

    In addition to this, with the Internet under its current governance, it’s still possible for people who want to communicate with much greater security and opacity to do so – provided they don’t use plaintext on a major social networking service.

    One of the best things we could do at this point in time is work to revive a project like Diaspora (https://joindiaspora.com/) and get its use to critical mass among our friends and family.

    What kind of record is it, and how would you access it? To replay it would take a lifetime, by definition. There is no memory, scrapbook or diary that contains everything, the very idea of biography is based on selection.

    “On the Impossibility of Drawing a Map of the Empire on a Scale of 1 to 1”, but for the quantified self.

    1. Replying to myself here, but right on cue, Malcolm Harris:

      But Twitter isn’t a map – it’s a territory. Rather than a megaphone or a flier, Twitter is a global city made of text.

      And each timeline a suburb …

    2. “On the Impossibility of Drawing a Map of the Empire on a Scale of 1 to 1”.

      That’s a lovely essay but as I recall it works on the idea of a physical map only. On a computer network in fact it would be exceedingly simple to build such a map. I see no theoretical reason why Google Earth (and Maps, Street View etc.) couldn’t some day be magnified on a scale of 1 to 1. Or even 2 to 1, etc. Likewise for a documented life. In fact, Gates’ idea of collecting data like body temperature and barometric pressure already suggests a surplus of knowledge on the part of the record (although there are also senses that a computer cannot map).

      1. Interesting. That’s not quite what I meant by the comparison, but yes—it would, perhaps, technically be possible to build an electronic mapping system so visually faithful that it appeared to mimic the “whole Empire” at a scale of 1 to 1 to a human viewer.

        I actually work in this field, so I’m in a position to give you some approximate values based on current technology. Bear with me, because this’ll get a bit kooky.

        Let’s suppose it was possible to gather remote sensed imagery of the Earth at a digital fidelity so high it was, to the naked eye, indistinguishable from looking at real objects at a normal viewing distance of about 8-10″—the “retina” concept.

        The fidelity of a “retina” display is about 300PPI, or ~120 pixels/cm. To put that in perspective, the tiles currently served by Google Maps are mostly based on aerial or satellite photography with a fidelity of (usually less than) 3 pixels/m.

        Lineally, that’s 4000 times as many pixels, meaning there are 16 000 000 times as many required for a similar map area in our “retina” map.

        Most existing consumer grade mapping systems use pre-rendered tilesets at various “zoom levels” to deliver maps, with PNG or JPEG tiles.

        Currently the “deepest zoom” on Google Maps is the 19th factor of two from Earth-as-two-tiles, which gives you tiles relatively faithfully representing the visual content of that approximate 3 pixels/m.

        To get to “retina” fidelity, you’d need log 4000 (base 2) extra levels: or about 12 additional zoom levels, the “deepest” of which would require 4^12 as many rendered tiles as zoom level 19.

        An estimated total storage cost for 19 fully rendered zoom levels of pre-rendered 256 * 256 JPEG tiles (assume 15kB per tile) for the earth’s surface (about 5 * 10^8 sqkm) is about 2750 terabytes (2.75 petabytes).

        That’s based on a very inefficient storage mechanism—with some modifications and filesystem customisation (you would need a novel filesystem technology to attempt this project, as the amount of data involved exceeds the theoretical maximum addressable by 64-bit architectures) perhaps you could get it down to 1 petabyte.

        With some strategic rendering on demand from highly compressed remote-sensed “retina” imagery of the earth’s surface, you could reduce it to 0.5 petabytes, probably, or even a bit less—so let’s use that as our figure.

        Now, what about this “retina” or visually faithful 1:1 scale?

        For zoom levels up to 31, you’d need a whole lot more. As above, it’d work out to about a factor of 16 000 000 more pixels to render.

        So that’d be about 8 000 000 petabytes, or 8 zettabytes.

        So it would require about 15-20 times the total amount of digitally stored information in the world today, to serve up a global-scale map at “retina” fidelity using technology similar to that behind Google Maps.

        There would be some considerable quirks in the user experience given that, rather than feeling as if one had a bird’s eye view, one would have a sense (when zoomed in) of crawling at floor height across a flattened representation of a surface.

        To counteract this, a new system of photographing oblique imagery would probably be required. All in all, the capture project for the data, and the methods used to visualise it in a satisfying way (think Google StreetView), would be more demanding than the mere storage—and would probably require additional data and computing infrastructure.

        1. That’s wonderful.

          “So that’d be about 8 000 000 petabytes, or 8 zettabytes.

          So it would require about 15-20 times the total amount of digitally stored information in the world today, to serve up a global-scale map at “retina” fidelity using technology similar to that behind Google Maps.”

          A yottabyte is 1,000 zettabytes, so the NSA data centre is… quite large, is it not? I wonder what Google’s global capacity is.

          1. I need to correct that assertion, actually—I was in full flight when I picked up the 0.5 ZB figure as the “total amount of digital information stored in the world today”.

            That figure’s actually from 2009, and I’d guess by now it’s probably about an order of magnitude out.

            The new NSA data centre near Salt Lake City is “designed to store data on the scale of yottabytes”, but “some published reports suggest” it has 5 ZB in initial planned capacity—so a lot less than a single YB at this point (although getting up towards the mark of my “map of the Empire”!).

            I quite liked Wikipedia’s pop science visualisation of a YB: “To store a yottabyte on terabyte sized hard drives would require a million city block size data-centers, as big as the states of Delaware and Rhode Island.”

Leave a Reply

This site uses Akismet to reduce spam. Learn how your comment data is processed.