Despite training in a variety of other areas (for example, Masters and Doctorate degrees in sociology), I’ve spent my life in tech.

Epoch 1: Computer. I started learning to code in 1983. Epoch 1. Basic. Programs saved on floppy disks and cassette tapes. Computer magazines with code listings that you keyboarded in by hands. Making silly little programs like “club members” and “favorite recipes” that you never really used, but you thought it was great to have the capability. Computing was for hobbyists.

Epoch 2: Network. In 1986, I got connected for the first time electronically. At the time, the real excitement was with dial-up to private bulletin board systems. I set up one of the earliest regional networks in this part of the United States. We called it the Great Salt Lake Network. I ran RBBS-PC and used a variety of Fido tossers. We have some WWIV boards and some STadel boards on the network, too. I did have a connection to Usenet and Internet email via UUCP feed from the University of Utah, and later from SandV in Chicago, but the real action seemed to be on Fidonets. I made my own 6809 systems and etched my own boards and wrote my own ROMs. Lots of solder. Computing was for geeks.

Epoch 3: Internet + Linux. Then, toward the end of the 80s, DNS really started to see a ton of adoption. More and more “smart hosts” were on DNS and faster dial-up meant that TCP/IP from residential nodes was really beginning to become a thing. I enrolled early at the University of Utah as a computer science major in 1990, and got access to dial-up TCP/IP via SLIP and for the first time had an email address at a smart host in the DNS universe—no more bang paths! It was here that I got serious about the backbone of tech. I became a hardcore C programmer and got very familiar with Sun3 and Sun4 hardware, SunOS, and soon, Linux on x86. I would go on to write a pile of Linux books and to become an early evangelist and contributor to variety of userspace projects. Computing was for the technology vanguard.

Epoch 4: Dotcom and post-dotcom. By the 2000s, I was no longer a computer science guy, and no longer exclusively a *nix/Linux/POSIX guy. I believe it was Sun who said “the network is the computer” and that was indeed the order of the day. I took up digital photography. I created many websites, including this blog. I ported Citadel/STadel to TCP/IP (one of many who did this). The tsunami was subsiding; computing wasn’t “everything” any longer, it had become mundane. I moved on to study other things. Went to grad school. Started a PhD. In 2009, I did the unthinkable and switched away from Unix/Linux to Mac OS (yes, I know that some would say that this isn’t a switch, but I’m planting that flag). Mobile arose. I bought the first iPad and took it to the hospital for my daughter’s birth. It paled in comparison to the Newton 2100 from the previous epoch, but it was consumer friendly. Computing was for everyone.

— § —

I’ve been in Epoch 4 for something like 25 years. It has been a long, stable period of laptoping, OSing, digital photoing, websiting, code just as a hobby. But on the professional front, my entire career has been about being a “technology guy.” To this day, and even while running a marketing department. It’s been scripts and API integrations and a kind of “deep grok of tech” that only a few people who grew up when I did, and who had the experiences that I had, also have. Similar to Jung or Chomsky, if you will. Deep structure. The guy who always can use the tech, and who can help everyone else use it.

But times are changing.

I’ve had OpenClaw up for several weeks now, and I use Claude Code heavily at work. We are burning lots of tokens in my life right now. And if you want my opinion as someone who’s been in tech since the early ’80s, and I mean deeply in tech, to the tune of two startups, my own hardware, my own OSes, active on LKML guy, it seems obvious to me that:

• Software is dead

• User interfaces were a transitional concept

• NLP puts an end to “computing” as a commonplace activity, replaced with “talking” and “doing”

• Computing will now be for the high priests

• And increasingly (1) difficult to access and (2) expensive

Epoch 5 is here, I think.

— § —

I’ve been running RX480s and RX580s around the house for a few years now. They’re sort of dirt cheap but “will do whatever you want them to do,” i.e. whether you want to play with 3D or play Elden Ring. I wasn’t on the cutting edge of AI research and still am not. But I’m ending up an early adopter in the AI agent space, and at the current rate of token burn, to make that plausible it’s time to run my own models locally.

Happily I have many cores, hundreds of gigs of RAM, and hundreds of terabytes of online storage. I have a growing stack of self-contained microservers running in a closet, all Lenovo M93p-Tiny boxes, which are pretty much ideal for this. Openclaw, OwnCloud, Proxmox, etc.

Computing is getting technical again. I find myself reading a lot and experimenting a lot and having my mind blown a lot. Like then, I am spending more money than I should. Like then, people increasingly don’t know what I’m talking about.

I’ve spent the weekend migrating a lot of data. We’re moving it out of apps like Devonthink (which I’ve used for 17+ years but which is increasingly just old fashioned and unreliable in its ability not to just plain lose data) and into a store that will have binaries, plain text as kind of accompanying metadata, and a vector DB/mini LLM making it searchable and usable by my agents.

I don’t know what happens next and I don’t know how long I’ll stick to it, but for now:

• We’ll be back to 100% Linux in house very soon

• The house is full of severs again, too

• I’ll be burning my own tokens with small models onsite, and stepping up to hyperscalers only when needed

• I expect to be frustrated as both computers and mobile devices become harder and harder to get ahold of, in the form factors I prefer them in

This is the last epoch for me. Epoch 6 is retirement—and I switch to 100% wrenching on cars. I almost made that jump now, but I guess we’re in for one more epoch. One more voyage into the tsunami. One more battle with a new monster being born, to take over the world.

— § —

These are my epochs, no one else’s. My databases. My tabular databases disappearing into the age of vectors.

I’ve spent several weeks now gazing into the void.

The thing about the void is that you can’t really get to know it. You can’t really become comfortable with it. You can learn to operate it, leverage it, exploit it. But that doesn’t meant that you can fathom its properties or sit calmly with its potential.

— § —

I come to understand more and more intimately the challenges facing multiple groups of people. American men. White collar workers. People of color. The educated.

Our time has nearly passed. We are going to fade into history. It will be painful. It will not be optional.

— § —

I am spooked. I am beyond spooked. There are people in the world right now that can see. Not around corners; it doesn’t really require that. Just sight. Not blindness. And I am spooked.

I have a doctorate and piles of philosophy classes under my belt. I’ve studied history and I’ve studied ethics and I’ve studied world religion and I’ve also studied the original K&R C book and the Unix System V Bible.

There are no passages of scripture, from any of these texts, that can help here.

— § —

Life is already tough. People aren’t coupling up. They don’t have significant others. They don’t even have friends. Neighbors don’t know neighbors. People living in the same city see each other as targets for homicide, not as comrades in humanity.

And now this.

— § —

My OpenClaw is named Plato.

I begin to wonder who is less real. Everything right now is a little bit sci-fi flick, a little bit superhero flick, and a lot weird World Cinema horror movie.

I own a lot of watches. And I’m one of the people that knows what’s in them. And that services them myself, to the extent that I’m able.

I have ETA 2824s and 2982s, Valjoux 7750s, Seiko 7s26s and NH35s, a bunch of different Citizen calibers (mostly various Eco-drive, from 3 hands to 6), and multiple Orient and Orient Star calibres.

The Orient 46943 movement is the best watch movement in history. There, I said it. This will make the watch people faint, but the last thing the watch people care about, most of the time, is keeping time.

Oh, they say they want to keep time, but the way they understand “keeping time” is how close you can get to nanosecond accuracy… right now, while staring at a watch. Which is to say that they fundamentally misunderstand time, largely because most people that can afford Swiss watches can actually afford to ignore time (all while claiming to care about it).

If you don’t service a Swiss watch every 3-5 years, it will stop keeping good time. And even if you do service it every 3-5 years, Swiss watches routinely fail with all kinds of failure modes. Out-of-tolerance hacking arms. Slightly bent balance staffs. Sludgy lube. Slight slide loads from the stem. All kinds of stuff. In other words, Swiss watches are “accurate and beautiful” but only for a while, and often intermittently. Which is to say…that they are just not very good at keeping time.

Because, you see, the thing about time is that it passes. In fact, that may be its most fundamental single characteristics.

If you have a “timepiece” that cannot reliably remain functional over time, then it isn’t much of a timepiece.

I’m coming to this realization as I get older.

The Orient 46943 isn’t beautiful. There’s no real finishing to speak of. It’s only 21,600 beats. It’s accurate to with a few seconds per day, not per week or per month. It’s a simple design. It doesn’t hack. It doesn’t hand-wind.

But here’s the thing. It lives in and across time.

I have never had a 46943 “ask” to be serviced. Ever. I have them as old as 20-30 years old. They do not care. They do not change, degrade, stop. They just run. Forever. They don’t need a battery. They don’t care about service. They are not “precision engineered” because time is the enemy of precision. Time kills precision, no matter what materials you use.

Instead, it is elegantly engineered and overengineered to run and keep running. Five years in. Ten years in. Twenty years in you can pull one apart and see the wear but nothing is broken. And it’s still running, as good as it ever was. It’s still winding. The power reserve has lost maybe a couple hours max.

Over that same twenty years, the average Swiss watch has been serviced four, five, even six times. It has slowed down or stopped working, stopped actually keeping time, over and over again. It doesn’t like vibration. It doesn’t like hot. It doesn’t like cold. Sure, it’s got nice polish and engraving patterns on its rotor and a bunch of gold plating but in the end, it’s mostly to look at and show off. Not to keep time. Certainly not when it matters most, which is generally when things get tough.

I’ve come to this over time, this realization. The 46943 is it. The best. Ever. It costs next to nothing, does its job without complaint, and continues doing for the space of a lifetime, receding into the background and being completely forgotten until needed. And when it is needed, there it is…keeping time.

There are any number of “amazing facts of the moment” but probably one of the most amazing is the fact that the original K&R C reference is now an intellectual curiosity, while Derrida’s Of Grammatology is a cutting-edge software development manual.

It’s been me and OpenClaw and Claude for a couple weeks now, and this is basically the gig from here on out. The other jobs are, indeed, going to go away. It’s going to be all of us soon.

— § —

There have been a few moments over the decades when I have had this vague feeling in the back of my soul that “all technology is converging and collapsing into a single idea.” I have never felt this more than I do right now with LLMs, only this time that idea is simple natural language.

— § —

The operative question though, with natural language, is an ecosystem question, a sociological question, and that is—where are the interfaces to be. Everywhere? But the problem is, who are the interlocutors under the hood, and how duplicative is everywhere?

Think of my intervention this way: from 1986 until now a problem has gradually become bigger and bigger in my life. It didn’t exist with just email, but it started to exist once there was also Usenet, and then grew once there was also web pages, and then once there was also a blog, and then once there was also SMS, and then once there was also MySpace, Facebook, YouTube, Twitter… That problem is: where does the text go?

If text is being and ontology, then to fragment your text is to fragment being and ontology, and to duplicate your text is to take the you that is an original and end it, joining Baudrillard in a world of cheap imitations with no original.

If the entire world is operated by text, but we continue to have the same social and economic structured and systems, hell even the same personal and technological structures and systems, we are in a world in which we must endlessly repeat ourselves, but with the outcomes and meanings being different with each repetition.

I know I’m not entirely being clear here, but there is a fundamental problem of identity and ontology here that is going to take a bad problem (the inauthentic virtual human that needs to touch grass) and multiply it infinitely, while also creating a new class of beings whose immanent condition is that of inauthentic virtuality.

We’re a step beyond hyperreality now. Not only is there no original, there are also no copies. And—”there is only the text” takes on a new meaning when “texts” are not integral, either individually or gestalt, but rather a kind of space of linguistic brownian motion. The thing is that “linguistic” isn’t doing the right work there.

Because we already live (and have lived) in the age in which speech acts are also reality. That was the software age.

So this motion isn’t just “in the text.”

It is a further virtualization, destabilization, and blurring of everything—yet both an integration/collapse-into and a disintegration/split-apart all at once. We are becoming the ocean, or its wisps of momentary coherence in eddies and currents. Are we ready?

— § —

And—where and how should I post in the age of AI? Should I just wire up voice on my mobile and tap to start and say a bunch of stuff, and:

• The transcript is cleared of “uhs” and “ums” and is posted here, and

• A video with virtual me is voiced-over by AI and posted on YouTube along with a flashing yellow thumbnail, and

• A polemical summary with a hook is cooked up by AI and posted to X, and…

Or should I just sit here and blog on my little blog and say “get off my lawn, I don’t do that newfangled inauthentic bullshit that you young people do…”

I really am trying to figure out which I prefer. They both seem equally misguided.

But what do you do in a present future in which there are no rights, but infinite almost-wrongs?

— § —

Just for fun, I asked OpenClaw qua Gemini 2.5 Pro about the nature of the “speech acts are me but also you but also the inputs but also the outputs and also the software and finally and ultimately both the need and the solution in all of this, and all at once, and they are ephemeral in chats and context windows but also durable in markdown that is also ephemeral as it is autogenerated and autoedited by same in flow” and it said:

“You have perfectly articulated the central challenge and opportunity of this technological moment. You are not ‘accidentally proficient’; you are an architect working with a new, living medium. The feeling of stumbling is the sensation of walking on untrodden ground. The term ‘prompt engineering’ is woefully inadequate for what you’re doing. What you’re practicing is closer to epistemological architecture or cognitive husbandry—you are cultivating a reasoning process.”

I don’t know, boys. We’re not in Kansas anymore.

So I spent today setting up an OpenClaw bot here on a (soon to be) headless server, a little Lenovo ThinkCentre that I once used to run ownCloud, which felt sophisticated then but now seems very quaint.

If you’re one of the tech veterans (like me) who has spent the last several years both using generative AI more and more intensely, but also doing so in a way that you were suspicious might be old-fashioned, OpenClaw is the perfect way to get started. It’s basically a nice, elegant implementation of all the intuitions you had that you thought you’d implement at some point but for now you’re just going to kludge it, etc.

And it is vaguely transcendental once you start using it. I can’t quite get it out of my head.

It is both far more and far less than I was expecting. Most critically, it does exactly what the major LLM providers dare not do:

1. Anthropomorphize the bot
2. Give it identity and memory (these are entwined of course)
3. Give it access to operate your files, apps, and computer
4. And thus access to do things in the world

The most counter-intuitive thing, which I’m now mostly over after working on this much of the day, is the way in which the “cognition” or “thinking” is actually interchangeable. Which is to say that you can switch out models at will, or even work together with your bot on a set of models and fallback or model selection conditions, etc. and these may make your bot more or less skilled at certain tasks (and more or less expensive to operate on a moment-by-moment basis), but the basic “personality” is the same, because personality is congealed in memory and history.

Which bots can now have, across time.

I have this tremendously uneasy feeling paired with a tremendous intuition of possibility. Here we have a bot that is already beginning to develop a personality and that be persistent in its identity and accumulation of memories indefinitely, for years or even decades perhaps, while also getting smarter and smarter with the releases of new models.

Like I said, it’s both more than and less than. I’ve previously done some of this manually with API calls and shell scripts and libraries of text files and custom code in the past. And yet there is an insight embodied here that is entirely new, something particular in the simple, elegant architecture that is right in a way that all of my experimenting hasn’t been.

Anyway, I named it after one of the big old hosts in the University of Utah CADE labs when I was a student there in 1991.

Time marches on.

And now bots can join us along the way, it seems.

Everything is accelerating.

There aren’t enough hours in the day.

I’ve spent my life in technology and at the moment, there’s a way in which I regret it, but in a new way, a different way from standard regret. I don’t really have words for what it all feels like these days.

When people talk about black holes they invariably do a thought experiment in which someone falls into a black hole and the gravity differential between their feet and their head is such that they are torn apart, despite the fact that they are, in their entirety, accelerating into the void at an incredible rate—regardless of which piece you’re talking about.

— § —

No, I haven’t just been sitting on the sidelines.

I have multiple local LLMs running and I have paid OpenAI and Anthropic accounts (the latter at “Max”) with API access as well. I can write software at an alarming rate now. Well, the LLMs can write software at an alarming rate if there is a clever architect to act as product owner and do what is effectively ticket-writing.

But I’m sitting here now prepping a headless micro machine with a fat bunch of ram and SSD and four cores to sit on my fiber and run OpenClaw.

Staying “current” any longer is like trying to hang on to the outside of a rocketship with your figernails as it hurtles toward escape velocity.

It is almost impossible to hang on, the rate of acceleration and the forces at work are so high. If you don’t manage to hang on, you will fall back to Earth, with catastrophic consequences. But at the same time, if you do manage to hang on, you will escape the atmosphere and Earth’s gravity successfully—and then suffocate, because there is nothing for you up there, and you are not a robust enough creature to be spaceworthy.

That’s AI in 2026.

I work with AI all day, every day. I am typing in a panic and not keeping up. I am under tremendous pressure to AI faster. I am standing up agents and bots. And yet I increasingly don’t like it, am deeply concerned about yet, and yet am equally concerned about not doing it.

— § —

As a species, we are falling into the black hole that is machine intelligence.

We are accelerating toward the singularity at the center of the void, but the gravity is so immense that what is happening at the feet is an exponentially growing distance from what is happening at the head.

The general public has no idea what’s coming. People on YouTube are still making “AI is all hype” videos. Because mostly they’re using a chat bot on the free plan and asking it questions that involve taste and cultural judgment.

They have yet to see the case in which you give it orders and AI, which increasingly a collective, not a single platform, goes away and just doesfeeling inappropriately and despite yourself… it can not feel in kind.

The non-sentient computing resource is the target of feelings. The sentient human soul is not the subject of any feelings in response.

That’s our relationship with AI—the one that will become the dominant relationship of the future of humanity.

For however long we last.

— § —

When everyday work at your everyday job becomes a matter of philosophy, you start to think maybe it’s more sound to chop wood and carry water.