Two Stories about the Computer (Pt. 1)

Hi everyone! I’m going to do something a little different this week. Instead of a deep dive into this or that specific aspect of AI, I’d like to lay out my broader sense of some of the cultural and market forces that gave rise to this moment. This is a story I've been putting together for the last couple of years; while deeply partial (in both senses of the word), I think it does go some of the way towards an account of why I think Silicon Valley was primed to go all-in on “AI,” and why the market was doomed to follow them. Next week, I’ll explain why I think that was a pretty silly gamble, in a way that might actually provide us a useful political opening. Have a good weekend!! – Theodora

Story 1 - The autohagiography of Silicon Valley

Here is one story that’s been told about the history of computers:

Upward the line of numbers takes its way

The invention of the personal computer, culminating in the smartphone, has been the most consequential technical innovation in human history. While there have been market adjustments here and there (the dot-com bubble, for example), it’s been essentially a story of continuous progress, with occasional leaps corresponding to new paradigms: mobile smartphones, cloud computing, etc.

On the whole, however, computer hardware and software have been improving linearly, if not exponentially, since the invention of transistor-based computing; they will continue to improve this way for as long as the policy landscape remains favorable.

Social media, delivery apps, constant connection and unprecedented access to relevant data have facilitated paradigm-shifting boosts in productivity, inaugurating a period of unimaginable and – assuming the government doesn’t crack down on the pioneering entrepreneurs driving the world forward – endless progress. If Silicon Valley founders are allowed to exercise their genius and innovate, this progress will continue.

AI is the next step of this progression, and the most important yet.

I hope it tasted good, at least

The ideas from this first story are unevenly accepted, but some of them have become some variety of common sense. The clearest statement of the maximalist position is probably the famous 2011 Wall Street Journal editorial “Why Software is Eating the World,” written by fascist venture capitalist Marc Andreesen; he’d later write the “Techno-Optimist Manifesto” exploring similar themes.¹

While libertarian techno-capitalism faces bipartisan opposition, and those of us who wrestle with buggy, irritating phone apps daily would be surprised to hear that they’re a step on the way to utopia, some of these ideas – such as the conflation of hardware progress with software progress – became essentially common sense.

Story 2 - The contingency of “progress”

Let’s start our second story about personal computers a little later than our first, in 2008:

ZIRPworld

In December of 2008, as a response to the Great Recession, the US Federal Reserve reduced the nominal interest rate to 0%.

Among other things, this period of zero interest-rate policy – abbreviated, hilariously, ZIRP – was a massive boon for venture capitalists like Andreesen, who specialize in risky gambles. When investors don’t have to worry about interest, they can take more chances and sustain periods of unprofitability for longer – they can always borrow more money to extend the gamble.

The dot-com bubble had burst almost a decade earlier, but between the creation of Amazon Web Services in 2002 and introduction of the first iPhone in 2007, the foundation was being laid for the next decade or so of Silicon Valley growth. ZIRP allowed venture capitalists to, fir example, subsidize phone-based services like Lyft (a prominent Andreesen-Horowitz holding) and Doordash – both of which were unprofitable for years.

It was only thanks to the unique zero-interest policy environment that these companies were to operate at a loss for so long. By picking risky, novel startups and bankrolling them, VC firms were able to secure “first-mover advantage,” the competitive advantage granted by being the first firm to secure a new market.²

Robbers, not demigods

But it’s Silicon Valley self-flattery to claim that it was pure first-mover advantage here. Founders of companies like Uber weren’t inventing new markets. They were undercutting providers in pre-existing markets, like taxi and delivery services, through smartphone software.

It was not a coincidence that workers in the sectors targeted by VC firms – car services in particular – often enjoyed stronger labor rights than others. Because interest rates were low, venture capitalists could subsidize these new app-based competitors. This allowed ride-share services to charge much lower prices than unionized taxi drivers, for example.

This wasn’t inevitable. In a better-regulated market, the engineers behind Uber could have been forced to work with pre-existing taxi services, engineering software that would make it possible to call and pay for a unionized cab ride from your phone. Gig-economic techno-scabbery would never have been tolerated in a policy and legal environment favorable to the basic rights of laborers.

The key advantage Silicon Valley had over pre-existing firms is that that – unlike, say, the cab companies – they had access to a bunch of experienced software developers. Lyft’s proprietary app would work a lot better on your phone than any competitor app cab companies could scramble to come up with. Combine that with the abundance of liquidity provided to venture capitalists by ZIRP and you have a recipe for sector-wide union-busting.³

By the time interest rates came back, venture capital had basically accomplished its monopolistic, anti-labor goals. While interested parties claim gig work as a libertarian victory for the masses (“set your own schedule!”), working conditions for gig workers were, and are, terrible.

By subsidizing services, VC allowed apps to gain users at artificially low prices, lock a customer base in, then ramp up the prices once the subsides went away. Apps like Uber and Lyft had essentially boxed out unionized cab services, locking users in just in time to jack up prices. Every restaurant had to deliver now just in order to be competitive, and they all contracted their deliveries out to app services, even as those services had to jack up their prices to make up for the subsides that are no longer available. Amazon ravaged brick-and-mortar booksellers; YouTube and streaming destabilized the union-heavy film industry.

The secret of SaaS

As a business model, app-based services could produce remarkable returns. This is because subscription-based software (“software as a service,” or SaaS) is, by-and-large, a remarkably profitable model. The expenses are largely upfront, in the initial development phase. Maintaining pre-existing software is much cheaper than building new software, and subscriptions mean the money keeps coming in.

But there were downsides. The ZIRP era had allowed growth of a kind that had become much harder to come by. People’s expectations had been distorted by the app boom. Shareholders wanted returns of a kind that even massively-profitable subscription services couldn’t provide.

And a lot of these companies seemed to underestimate how expensive it would actually be to maintain their services. Take film/TV streaming: the original promise of Netflix’s streaming services – any movie, ever, for $8.99 a month – was delusionally optimistic, especially in the movie industry, which had some of the strongest unions in the US.

This is, in part, thanks to perhaps the foundational sleight-of-hand performed by the modern software industry – the idea that computing is basically non-physical at this point, and therefore more-or-less free: a “cloud,” floating breezily above the terrestrial earth. But it’s not. As AI has proven again and again: for as small as personal computers have gotten, the heavy-duty ones take a lot of space and cost a lot of money. (More on this in part two.)

And so begins the process Cory Doctorow calls “enshittification” – making the software worse and worse in order to ensure returns for shareholders. Ads become ubiquitous. Algorithms get more and more annoying. You see less of what you want to see and more of what will get them ad money. (It’s worth noting that, apart from a few major exceptions like Amazon, big tech basically runs on ad money.) Advertisers themselves start to get screwed – as best as anyone can tell, it seems like Meta in particular – 98% of whose revenue is in ads – is probably guilty of rampant ad fraud.

Streaming services hike their prices, then break into a Protean complex of separate services, all of which have nineteen available subscription tiers, eighteen of which show ads. It’s all falling apart, it all sucks, nobody likes it.

This brings our story up to around the start of 2020, an uneventful year that will kick off a placid, reflective decade. Just kidding!

To be continued…

Oh no! How, pray tell, will ordinary red-blooded American venture capitalists like us survive the existential threat of Joe Biden’s Communist dystopia? Stay tuned for the follow-up later this week, when I reveal that the answer is, apparently, “throw literally two trillion United States dollars at janky chatbots and unbuilt data centers.”

The Genuine Intelligence Project is an initiative from the North Carolina branch of the American Association of University Professors. Check out our website, and follow us on Instagram and BlueSky!