the return of big computing

This may be an appropriate article to begin a new feature on varnelis.net. A few times a month (if that), I'll be sending out more lengthy or provocative posts via email in hopes of stimulating more discussion and to avoid the perennial emails to my colleagues suggesting that they go read my blog. If you're reading this post that way, it's because you're a member of my "inner circle." If not, and you want to be, try signing up in the block on the right of the page. Contact me if it doesn't work.

In any event…

Vijay Patel sent me a link to this article by Nick Carr in the Guardian: Software Companies are Building Their Way to a Very Material Future. In the article, Carr points to a new building boom among computing firms such as Google or Microsoft. Remember Manuel de Landa's thesis that meshworks lead to hierarchies and hierarchies lead to meshworks? This is perfect evidence of the case, as these large corporations require ever bigger data centers for their massive server farms.

microsoft data center under construction

From the 1970s onward, the paradigm of big computing broke down as more and more computational power came to reside in personal computers. Many readers of this blog are probably too young to remember those pre-web days when you would go to a lab to connect up to a mainframe via like a DEC PDP-11 over a dumb terminal; my first experience computing, in fact, was over an ASR-33 teletype. Eventually, as the story of joyous liberation goes, power was distributed to the PC.

But it's a myth to think that the Net is distributed, either physically or somehow inherently, in its ideology.

Don't mistake network culture for distributed culture. Far from it.*

The migration of the means of production back up to the web and the growth of social networking sites is generating a mass wave of consolidation and aggregation. As the Internet has grown, so has the share of Internet traffic dominated by the top ten web sites (see this article by Richard McManus). We may be seeing a thousand flowers bloom at the level of content production, but control is in the hands of the few.

And now, with Web 2.0 taking hold, we have software applications migrating to the Web. This is attractive for individuals since licenses are often free in exchange for precious demographic data that would otherwise be unavailable—hence no more need to pirate—and to corporations since annual leases can be more conveniently written off than outright purchases.

But it also suggests that after three decades of the means of production drifting downward into our hands, they are beginning to slip away from us again. The meshworks are breeding new hierarchies.

I don't see this as a positive step. I also don't hold much faith that agitation on the part of the anarcho-libertarian streak on the Internet is going to change matters much. Rather, as network culture develops, we find the coils of the serpent that is capital tightening again. Indeed, this seems to be a salient feature of network culture: that if big computing 1.0 was Fordist, and the PC was post-Fordist, then big computing 2.0 is a hallmark of network culture. In future posts, I hope to explore what this means.

*Some day I should come up with a list of my key theses. This should be one of them.

This may be an appropriate article to begin a new feature on varnelis.net. A few times a month (if that), I'll be sending out more lengthy or provocative posts via email in hopes of stimulating more discussion and to avoid the perennial emails to my colleagues suggesting that they go read my blog. If you're reading this post that way, it's because you're a member of my "inner circle." If not, and you want to be, try signing up in the block on the right of the page. Contact me if it doesn't work.

In any event…

Vijay Patel sent me a link to this article by Nick Carr in the Guardian: Software Companies are Building Their Way to a Very Material Future. In the article, Carr points to a new building boom among computing firms such as Google or Microsoft. Remember Manuel de Landa's thesis that meshworks lead to hierarchies and hierarchies lead to meshworks? This is perfect evidence of the case, as these large corporations require ever bigger data centers for their massive server farms.

microsoft data center under construction

From the 1970s onward, the paradigm of big computing broke down as more and more computational power came to reside in personal computers. Many readers of this blog are probably too young to remember those pre-web days when you would go to a lab to connect up to a mainframe via like a DEC PDP-11 over a dumb terminal; my first experience computing, in fact, was over an ASR-33 teletype. Eventually, as the story of joyous liberation goes, power was distributed to the PC.

But it's a myth to think that the Net is distributed, either physically or somehow inherently, in its ideology.

Don't mistake network culture for distributed culture. Far from it.*

The migration of the means of production back up to the web and the growth of social networking sites is generating a mass wave of consolidation and aggregation. As the Internet has grown, so has the share of Internet traffic dominated by the top ten web sites (see this article by Richard McManus). We may be seeing a thousand flowers bloom at the level of content production, but control is in the hands of the few.

And now, with Web 2.0 taking hold, we have software applications migrating to the Web. This is attractive for individuals since licenses are often free in exchange for precious demographic data that would otherwise be unavailable—hence no more need to pirate—and to corporations since annual leases can be more conveniently written off than outright purchases.

But it also suggests that after three decades of the means of production drifting downward into our hands, they are beginning to slip away from us again. The meshworks are breeding new hierarchies.

I don't see this as a positive step. I also don't hold much faith that agitation on the part of the anarcho-libertarian streak on the Internet is going to change matters much. Rather, as network culture develops, we find the coils of the serpent that is capital tightening again. Indeed, this seems to be a salient feature of network culture: that if big computing 1.0 was Fordist, and the PC was post-Fordist, then big computing 2.0 is a hallmark of network culture. In future posts, I hope to explore what this means.

*Some day I should come up with a list of my key theses. This should be one of them.

Posted in Uncategorized

2 thoughts on “the return of big computing

  1. Mega-centers won’t build monopoly power
    I not only remember the PDP-11 but my terminal for it was a DecWriter. I’m old enough to have also bootstrapped a PDP-8, though admittedly I was a teenager at the time.

    Like everyone else I’ve read the articles and seen the pictures of the immense data centers being built, but I can’t agree with your assessment that these mega-centers give them an unfair advantage going forward. I’m biased because I work in this space as co-founder of 3tera, but utility computing services like EC2 and The Gridlayer already enable anyone to harness hundreds or even thousands of servers to power their applications without any capital expenditure or long term contract.

    1. But as I suggested, compared

      But as I suggested, compared to a decade ago, the Internet already is more centralized. Take search engines… I used to visit alta vista, lycos, yahoo, hotwired, northernlight and other engines regularly. Now I visit Google.

      Thanks to RSS, I am able to keep up with about 100 blogs in my daily feed—that certainly is a big improvement, although if a clear leader emerges in RSS aggregators, I will likely be doing that through one site rather than through a firefox plug-in—but what I'm suggesting in the above post is that after a certain point, decentralized structures breed centralized structures. If things go well, the latter will breed decentralized structures again. If things go badly, after a time of extreme misery, the latter will disintegrate.

Comments are closed.