Cozytech

Towards a viable alt-tech philosophy made up of individually delusional, collectively rational parts

A few months ago, I came across Tailscale, and a competitor, Zerotier. These are similar products that let you run secure, private networks that do more sophisticated things than traditional VPNs, and do so in two interesting and differently opinionated ways. Neither is a consumer-grade product. They are meant for people with some system administration skills, so you have to know something about networking tech to use either. Both also clearly have a different DNA than typical Silicon Valley products. Tailscale was co-founded by Avery Pennarun, who is a well-known tech blogger of the curmudgeonly sort, under the handle apenwarr, and under his real name he has written a manifesto for the vision driving Tailscale, The New Internet. As with anything written by curmudgeonly alt-techies, it is stimulating, insightful, and original, and you don’t have to agree with all the details or even the overall thrust of the vision to get a lot out of it. Zerotier too, has an interesting philosophical origin story. Here is an early blog post by founder Adam Ierymenko. In this picture, we also have Wireguard, an open-source project with similar philosophical underpinnings.

I found my reaction to The New Internet to be divided. It is the sort of idea that reads delusional if you take it as a vision for replacing the internet as we know it wholesale, but quite rational if you think of it as one piece of an emerging puzzle, where other pieces balance it out via various oppositions. Tailscale, understood via its guiding vision of the “New Internet,” has what I think of as an individually delusional, collectively rational character. It makes sense as part of certain alternative tech ecologies that may emerge, but not as a totalizing vision of a “New Internet,” or as part of the prevailing Tech ecology. Similar things could be said of Zerotier and Wireguard.

Tailscale, Zerotier, Wireguard, and the philosophies driving them, belong on an emerging alt-tech landscape I want to talk about today. I am going to label this landscape cozytech. Cozytech isn’t a prescription or a description, or a singular subculture. It is, rather, an interesting reachable equilibrium that a whole constellation of alt-tech subcultures, each individually perhaps delusional, but collectively rational, are driving towards, and might actually reach. If we get there, we might actually have a muscular alternative to Silicon Valley that doesn’t feel like an insipid and quixotic political-resistance theater.

But let’s set the stage here.

The word technology today has become synonymous with a certain flavor we’ve come to refer to simply as Tech (I capitalize the term) epitomized in our time by the Silicon Valley approach to building things. But the Silicon Valley flavor is only one chapter in the history of Tech. The core element is the presence of an organized landscape around a key “eating” element that can colonize everything. As I wrote in my June 7, 2019 newsletter issue, Can Tech Die?:

In the last 400 or so years, there’s always been a relatively active “eating” tech at work in the world, being driven by the technological imagination, limited only by how clever people are capable of being, and driving the history of the world faster than any other competing force.

When such an element exists — and software has been the element for the last 30 years — “Tech” dominates all of technology. When there isn’t, technology becomes a more diffuse, less structured background force. During such periods, Tech lies dormant, while technology itself continues to evolve in a different mode, defined by an emergent equilibrium of individually delusional, collectively rational parts. Cozytech is my name for one such era that might possibly be emerging.

This is a punctuated equilibrium model of technological evolution, and “cozytech” is my proposed characterization of a possible next equilibrium, past the punctuation discontinuity that is Silicon Valley Tech.

We get capital-T Tech during Cambrian-explosion style events, and then just ordinary background technological evolution (I’ll shorten this to “background tech”) in the periods between the explosions. During Tech epochs, to use Brian Arthur’s terminology in Nature of Technology, a major natural phenomenon is typically discovered and harnessed in broad ways, though this is not always the case.¹ During Tech epochs, background tech dynamics get suppressed or even entirely squashed through various mechanisms — getting starved of resources and talents, Tech offering cheaper/more convenient solutions to problems that were the preserve of background tech, active political oppression, enclosure and capture, cronyist corporatism, and so on.

But like Tech, background tech too never completely dies out. Even during peak Tech eras, it persists as an endemic phenomenon on the margins, ignored or perhaps laughed at, waiting for the next opportunity to break out, shrug off the waning power of Tech to suppress it, and take over the logic of the narrative.

We can distinguish several Tech epochs in history besides Silicon Valley — interchangeable parts mass manufacturing (late 19th century, 1870-1910 or so), steam engine (early industrial revolution, 1770-1825) and printing press (mid 16th century). I think Tech, as such, requires 4 conditions to exist:

1. Major physical natural phenomenon being harnessed OR serendipitous convergence of several minor ones.

2. Systematization of key bodies of knowledge beyond trial-and-error tinkering and savant-ish engineering skill, allowing for predictable return on investments in the form of a landscape of investible opportunities staffed by concentrated pools of mediocre engineering and scientific labor, with little or no need for savants.

3. Political-economic conditions with enough surplus and stability for intelligent capital markets to coalesce, in the form of effective institutions (DARPA, VC sector, Medici family, Softbank, Mohammad bin Salman’s regime, Chinese emperor, Xi Jinpeng, whatever) that can act on those opportunities.

4. Cultural conditions allowing for a compounding path-dependent improvement process to unfold for at least several generations at a stretch (true in early industrial England and Silicon Valley, but not true in medieval China for eg) — this is basically Joel Mokyr’s model.

Tech requires a rather remarkable constellation of forces converging to create a larger phenomenon, so we should not expect it to happen very often or sustain for very long. If even one of the conditions fails enough, Tech as an explosive evolutionary epoch cannot persist. Instead of a Cambrian explosion punctuation regime, you get the slow background evolutionary type of epoch. By my rough accounting, before the Neolithic revolution, you had 100,000+ years of pure background tech evolution that crept along. After that revolution, circa 10,000 BC, I’d say we’ve had about 20% of history being “Tech” regimes, and 80% being “background tech” regimes. That 20% of Tech probably created 80% of the “Progress,” but the 80% of background tech, in delivering the remaining 20% of Progress, probably locked down the gains in deep ways. In the last 400 years, the Tech epochs have been getting longer, and the background tech eras have been getting shorter, but I think this process has a limit. We’re not going to get to a continuous Tech boom with no background tech interruptions.

It is these background regimes I want to talk about. What happens when one or more of the conditions fails? How do we transition from an active Tech epoch to a background technology epoch?

I think we’re in such a transition right now:

1. The harnessed force condition is failing (Moore’s Law troubles). There’s at most another 10 years and maybe 40x improvement headroom left, and then the music stops unless experimental post-lithographic technologies succeed.

2. Systematization is starting to break because the major knowledge frontiers we’re pinning our hopes on (AI, energy tech, biology, terraforming) resist strong systematization and depend on a grind of trial-and-error that works better in background ways. There’s still science as a potent force, but not as a high-leverage one. You can’t mindlessly through big lumps of capital at huddled masses of engineers and expect 10x returns reliably.

3. For the moment, political-economic conditions are still positive. There is plenty of surplus capital in the world desperately looking for places to go, though with unrealistic expectations of returns. And despite widespread institutional decay, contrary to what the burn-it-down crowd believes, there is still capacity for effective organizing around opportunities, even if in unfamiliar forms. But we’re getting there. Institutional capacity decay is real, and if things continue as they’re going, we will soon be unable to organize effectively beyond very small scales.

4. Cultural conditions are arguably broken, as I argued in Silicon Valley Vibe Shift. Silicon Valley is no longer a good environment for creating multi-generational path-dependent compounding results for a variety of reasons, the growing illiberalism being a big one. Visible signs include the shifting of initiative to the larger platform monopolies, enshittification of those platforms, and disenchantment with the region in the traditional entrepreneurial class. The specific SV formula of a particular horizon of investment, an approach to individual wealth-building through stocks, “creative monopoly” business models, and an “aggregation theory” approach to scaling, is breaking. More money is being raised and distributed than ever, but it’s something of an illusion because the underlying model is morphing in unrecognizable expedient ways (blurring into private equity for instance, and getting increasingly dependent on protectionist cronyism and state patronage, though the free market rhetoric continues stridently). There are no other credible hotspots with alternative cultural conditions.

It doesn’t matter if I’m right or wrong on any of these assessments. I just have to be right enough on any one of the four, for the current Tech era to break. Intense evolutionary bursts are rather fragile phenomena. The bad news is that the so-called “Dark Ages” between them tend to last at least a few decades on average. The good news, as I argued in Can Tech Die, is that Tech doesn’t die, and we’ll almost certainly see another evolutionary burst within a few decades, within our lifetimes. But a few decades is enough to kill a particular institutional landscape/manifestation of Tech, even if the Cthulhu beneath it lives on. “That is not dead which can eternal lie, with strange aeons even death may die” etc.

But the even better news is that “Dark” ages are not actually dark, and that background tech epochs are not just highly valuable by their own internal logic (which we’ll get to), they’re actually necessary for the health of the overall yin-yang of foreground Tech and background tech. Trying to have only one of the two is like trying to be awake 24/7 or trying to sleep 24/7. You need a healthy alternation of sleep and wakefulness.

We’re entering a background tech epoch right now, and it might be healthy or not depending on how exactly it evolves. Will we get the slow improvement mode of the European Middle Ages, which saw steady advances across the board in things like metallurgy, agricultural science, and so on? Or will we see a long stagnation of the sort many fear?

I think there are many bad equilibria background tech might end up in, but at least one good one that I’ve started thinking of as “cozytech.” In this essay, I want to sketch out a preliminary view of the cozytech equilibrium. It builds on my earlier socio-cultural notion of the cozyweb, which, despite being the one to coin the term, I kinda dislike and resist because it seems like a mode of fearful retreat to me. But if you add better cozytech to the cozyweb, maybe things will start cooking in interesting ways, and we’ll get a mode of bold advance rather than fearful retreat.

Maybe instead of being a place of fear, the Dark Forest can be a place of quiet, peaceful, background evolution. An era of locking in the gains of the waning Tech era, and working on raising the floor, after the era of raising the ceiling we’ve been in for 40 years.

So what is the cozytech equilibrium, how do we start herding ourselves towards it, and how can it begin to catalyze some good chemistry with the cozyweb that is already in place?