IT Futurology and the Terabyte iPod

Sun’s just wrapped-up its Customer Engineering (CEC 2007) conference in Las Vegas; for obvious reasons I was not able to attend, but several weeks ago I had committed to present:

IT Futurology

What will happen when the MySpace generation grows-up? When employees want Facebook rather than a phonebook? When your monthly report is on your iPhone, your spreadsheet’s on your Wiki, your e-mail’s moved to Google, you haven’t got a home directory (or a PC) any more, and blogging’s no longer a buzzword – it’s just what everybody does to stay employed?

If these questions aren’t on your mind, then they should be. Sun’s internal culture is three to five years ahead of the market, and even we’re running to keep up with the pace of change; if that’s *our* challenge then imagine what will happen when investment banks want to hire top talent five years hence? Employees will vote with their feet, and staff retention will be proportionate to environments which let staff do what they want, how they want, and Sarbanes-Oxley may no longer be enough of a lever to control their behaviour. Will your customer’s predicament become a matter of “do what your staff want, or die?”

So rather than cancel the whole thing, I thought I would try something different – another Lessig-like presentation, though without an audience which did make the process somewhat harder; plus I was really pushed to find time for script generation and recording amongst all the other stuff I had to do – so it’s not quite come out the way I wanted, more texty rather than visual – but it’s good enough for a wider audience and I would prefer not to delude myself that I will “re-record it and publish a better version, later” without first getting some feedback on the content.

The result is now available on blip.tv; it runs for almost 40 minutes and is will be obviously meant for an audience of techies who make hardware and software, but the ideas – including what you could do with a 1 Terabyte iPod, or how you could do away with the intermediation of a web server – should be of interest to quite a lot of folk.

If I had the chance to do it again the talk would be a little different, I would like to record the audio with a small audience (or with an audience of more than just Bart Blanquart, who helped me greatly but alas does not constitute a laugh track) – and then host a discussion to open up more ideas and possibilities.

Maybe next time. Herewith the video:

It’s probably nicest to play the video directly on the blip.tv page (provides for a bigger video / better graphics) but the embedded player version above should work OK.

21 Replies to “IT Futurology and the Terabyte iPod”

  1. Pingback: c0t0d0s0.org
  2. Alec, thank you for posting this presentation! I enjoyed it really much. In fact, you’ve created an embodiment of much that I was thinking about for a long time.
    Actually, once your TBPod will reach a critical mass of adoption, it will start creating its own peer-to-peer network, eliminating traditional ISPs and mobile phone providers. You can see a predecessor today with fon.com: What happens if their hot spots reach critical mass and start talking to each other rather than to ISPs?
    But the future wouldn’t be worthy of history if it didn’t continue in cycles. People will start introducing data centers again, to “improve the TBPod service” and rent their services to the TBPod user at a small fee so they can access their data faster/more reliably/etc. and then we’ll see a shift back to huge server installations and so forth.
    Anyway, thank you for a most excellent wake up call to many people out there!

    Cheers,
    Constantin

  3. Hmm… interesting ideas.

    However, I can see that this will only really work for “general computing,” i.e. where the data size is relatively small (i.e. less than a TB), at least in the first instance, and where the objects are also small (i.e. only in the 10s of GB or smaller) and that the requirements of the usage are not CPU intensive.

    The above conditions definitely do not co-incide with either data intensive operations or CPU intensive operations (HPC etc.). The data cloud paradigm doesn’t fit the application.

    It will also require that the density of devices is such that the devices will actually be interconnected often and that the Internet is pretty well all pervasive and fast.

    These conditions are likely to be available within metropolitan areas in advanced nations but as soon as you remove yourself from those areas you exceed the limits of the data cloud. Unfortunately, even in the case of the most positive predictions about accessability coming true, the majority of the planet will not be within that data cloud. When you’re disconnected from the cloud you’re potentially disconnected from your data.

    One brake I can see on this vision, at least as far as commercial organisations and government use is concerned is the fact that confidential and/or secret data will be held on equipment that is not controlled by theose entities. It maters not that the data is encrypted as the organisations involved won’t trust the strength of the encryption however good you say it is. This would be especially true for governmental organisations. After all, all it needs is an enemy to secretly crack the encryption technology and/or brute force it, or poison the data cloud with data which SEEMS to be the real data but is garbage (hence performing a denial of access to vital data) and it all falls apart.

    I’m not sure that the potential benefits would outweigh the deficits for this technology, other than purely for the consumer level or for the sort of documents which are purely personal.

  4. Alec, I enjoyed your video presentation for the CEC. You’re so cool. While I enjoy people guessing about the future, I always enjoy it more when they turn out to be way off. Hopefully your prognostications, if we can call it that, are close to future realities. Anyways, I hope that you review your presentation again in 2012 and we’ll see what happened (or didn’t).

    I was playing with Rinda in Ruby a few weeks ago for a message passing part of my research project. (Rinda is a tuplespace /
    objectspace bundled in Ruby.) I played with JavaSpaces before and looked at the Linda research work previously. (I find it interesting that years of research and millions in funding gets boiled down to about 1100 lines of code and embedded docs bundled in a scripting language library.)

    It dawned on me that having my own personal globally-accessible and
    highly available objectspace for my own objects would be really useful.
    I thought about the business model for such a service and dismissed it
    rather quickly as too expensive in terms of bandwidth and storage —
    from a “centralized” service provider point of view it is. However as
    you point out, dividing it up and spreading it around makes it a much more interesting and potentially cost-effective way of doing it. The
    paradigm shift around the applications that could run on such a network
    persistent storage service would be neat to see.

    Anyways, loved the talk. Thanks for making it available. Some thought-provoking views.

    Keith

  5. Pingback: CentronX' Blog
  6. Hi Alec, great presentation!

    I’ve been working in the quiet for the last year researching my own view of what you describe in the video.

    I have a question I’ve been grappling with for some time now to do with your slide regarding blending with the Internet and not attempting to encapsulate it. I was wondering if you could elaborate on your thoughts there?

    I ask because I’m of two minds at the moment in that regard. Brought about mainly by the imperative nature of the current architecture as a whole – and what I see as problems with regard to the user tools required in order to scale into this kind of distributed model. I’m torn by thoughts necessitating a deterministic approach to get where I want to go. With FPGAs and the like entering the high performance computing arena now as augmenting processors, it only exacerbates my thoughts. Mobile devices practically are FPGAs with other peripherals plugged into them. What with tools such as the Eridon UnifiedLogic now coming online, I’m seeing the separation of hardware and software beginning to blur. If I think even further ahead; dynamic devices such as Element CXIs Elemental Computing Array just blur that further. I’m especially torn when it comes to operating systems and maintaining legacy usage there. I keep wondering whether or not distributing these through encapsulation is ultimately where we’re headed. Perhaps I’m thinking too far ahead and that this will be a second coming after teraPods… why I’d appreciate your thoughts on the matter. As I see it now web browsers are already legacy applications. They themselves are not distributed. The hurdle however, and one I keep coming up against – is secure distributed application execution on untrusted hosts. I don’t particularly like the trusted compute agreement models I have thus far. That may only get me TeraPods. ;^)

    If we’re to spread technology further into our would as we are already doing, those ultra-thin clients will need to just be about communication when you get to particle level of paint on a wall. Have you ever read Only Forward by Michael Marshall Smith? 🙂

  7. > So you have discovered dunkertons cider, 15yrs after me.

    Valdis, dude! I discovered it a couple of years after they opened, but as for having a stock of it around the house, yes indeed… 🙂

    Hugs!

Leave a Reply

Your email address will not be published. Required fields are marked *