When I first started blogging, the whole thing was a series of shell scripts running out of ~/bin on my PC.

I’d type “post” at the command line and it would use a quick and dirty heredoc block to enable me to just pound in a post right into the terminal window. I’d end with EOF, the script would use sed and awk and so on to update other links in adjacent files to point to the new file (for top of blog, next/prev stuff, and so on), and then would FTP all the relevant files up into a static folder.

At some point, I adopted Graymatter. I think this was maybe in 2002 or so, though it’s all getting fuzzy a quarter century later. (Graymatter was a CMS written in perl. Yes, really.)

From that point forward, posting was a matter of visiting an admin backend URL on the live site, logging/authing in, and using a web-based interface to create a post.

— § —

All of that is fine in theory, but in practice, at times it has meant less posting. Sometimes having to complete the act of visiting the backend URL, logging in, then navigating a posting UI is just a step or two too far; instead, things don’t get posted.

Like everyone, I’ve adopted Obsidian this year and of course am building out AI agents everywhere. So for the first time in a good long time, I have a workflow again that posts without having to navigate the web UI, effectively syncing an Obsidian folder to the CMS, meaning that theoretically i can just pound stuff out in Obsidian and it will end up online.

Whether or not it works (technically) remains to be seen.

Whether or not it works (for reducing friction and changing habits) remains to be seen.

But it’s worth mentioning because I’ve been meaning to implement this ever since I started using Evernote in like 2006, and through every successive notes and everything-bucket style app.

Now in the age of agents, it can finally happen, because you can just throw an agent on it, rather than having to dedicate an hour to coding and another hour to debugging.

— § —

Meanwhile, I have a CT scan in a few days. I hope things are not actually brewing, but they may be brewing. I am taking stock, I guess.

I get into these funks where I’m not answering the phone. Not because I don’t want to, exactly, but because I just feel like I don’t have the energy to talk to anyone or to think about anything.

This year I am operating at the edge of my ability to maintain structural integrity, and sometimes it feels like one more conversation and the whole thing will blow up. I don’t know what blow up means in this context, so no point asking.

— § —

I don’t even know what to post here. All night I’ve been having fragments of posts and the impulse to post float through my head. Now that I’m here at a screen typing, all I want is to get away from it.

— § —

I don’t know exactly what’s wrong, but something’s wrong. Something needs to be addressed. You feel isolated but you won’t talk to your friends. You feel unheard but you don’t want to express yourself. You feel stuck but you don’t want to change anything.

I don’t know what that is. I’m sure there’s a psychologist or ten who would have a name for it. I’m also sure their advice would just irritate the f**k out of me rather than helping at all.

I joined Gmail during the beta in 2004, when it was by invitation only. I fought hard for my invitation, combing my network until I found someone in the tech publishing world that was willing to send me one.

A full gigabyte for email. For free! Imagine! Obviously it would never fill up!

— § —

I’ve been a Google shop over here for a long time. As an OSS contributor and evangelist, their first motto, “Don’t be evil,” resonated with me, and as time went on and they offered more services, I adopted more services.

Fast forward to the present day and I use Gmail, Google Calendar, Google Docs, Google Drive, Google Voice, Google Photos, Google Maps, Chrome, Chrome OS, Android, and Google Fiber and have for a long time. There isn’t a single item in that list I’ve been using for less than a decade. I even pay Google for extra storage. While they existed, I also used Google Wave and Google Plus.

Even when I realized that things weren’t going to last forever, when they changed their motto away from “Don’t be evil” (has there ever been a more “they’ll tell you who they are” move in all of business history?) I stuck with Google stuff, if only as the result of inertia.

I mean, my main Gmail box, that one that goes all the way back to 2004, has been used through two graduate degrees, multiple books written, and many freelance and work projects. It’s subscribed to mailing lists and is a kind of archive of my adult life. It’s got nearly a million messages in it (hence me paying for extra storage).

— § —

Today I opened a Fastmail account. I’ve been moving other parts of my life to AWS.

The sale of Google Fiber and the end of support for “send as” in Gmail are just it for me. I’ve had enough. I have multiple email addresses going back years, including multiple .edu addresses, and I receive on them all and don’t plan to stop. If Google was going to allow us to provide an alternate SMTP service, I probably would have stuck with it using a Sendgrid plan or something like that.

But no. That’s not how Google does things. They disappear things. Like they did with Wave and Plus. Like they’re doing with Chrome OS. And yes, they will still allow some support if you upgrade to Workspace, but it’s not cost effective for what I need and they don’t provide any migration options for Gmail-to-Gmail, ironically. So I’d have to nurse along a huge export and reimport using probably IMAP. Obviously an LLM could just take care of it for me, but at a million messages, that’s probably many days of throttling and fighting.

I haven’t liked their UX in a long time, but now that I also have to find other providers for things anyway, it’s just time.

— § —

I think Google is dying.

Obviously there’s the fact that they’re having quarters in the red and their search business is no longer the dominant force in the tech economy. But there are other signs. Like selling off Fiber. Like enshittifying Gmail. Like all the extra crap they increasingly add to Android themselves.

They’re not just in that part of the cycle where they enshittify, they’re in that part of the cycle where they enshittify because think these things aren’t worth doing any longer, so they’re just going to punish anyone still using them.

I get it. They want to focus on Cloud and AI.

Only their AI is enterprise only via Vertex/Gemini Enterprise. Yeah, okay, they claim to have a consumer offering, but it sucks. I should have been an ideal Gemini bundle customer but I stopped using it soon after trying it because Anthropic and OpenAI were both light years ahead, and that’s still the case. There were seven months between 3.1 Pro and 4 Argon. I haven’t tried Argon yet, but Google’s just too late. Under no circumstances should I have better inference *under my office desk* than I can pay for at Google, and yet that’s been the case for months.

Qwen 3.5 122B was better behaved and more capable for agentic than Gemini from day one, and Qwen 3.8 Flash Next put every Gemini 3.x release absolutely to shame. The fact that I didn’t even realize that Argon had been released until tonight (days later) pretty much tells me what I need to know about it.

Meanwhile, Google’s claiming there are massive Cloud gains, but nobody in the industry takes Google Cloud seriously because it’s an unstable, fractious mess. The growth they’re “seeing” comes from a drug-pusher approach. They offered thousands in free Cloud services credits to anyone willing to build things in Cloud this year, getting smaller orgs and devs hooked on Google services in the short term by virtue of the fact that they built in Google Cloud to take advantage of the fact that Google was giving tons of it away. But they won’t stay long.

I even went and got me a few thousand in Google Cloud credits but then ended up not using them because the only thing I really have cared to work on this year that would have been of interest was AI, Gemini 3.x wasn’t even useful with thousands in free APi credits. It couldn’t code reliably or use tools reliably, and it hallucinated on almost every turn.

— § —

Anyway, who knows, maybe Google isn’t dying and I’m just wrong. But it’s dying to me. I’m switching out of Gmail. Which probably implies a switch away from Calendar and Drive as well. I’ve played with OwnCloud some and I suspect that’s where we’re going to go. Another ThinkCentre m93p Tiny to stack up and put to work.

The funny thing is that I’d happily pay Google for things, but they either don’t have things I want to buy (in the case of AI) or they don’t want to sell what I’m willing to pay for (in the case of send-as capable Gmail accounts and fiber to my dwelling).

So now I’m paying Fastmail and AWS (and probably soon others besides).

Google is doing what IBM once did. Decide it doesn’t want consumers’ money and is going to focus on enterprise. Which basically means it’s already an also-ran, the next IBM.

Anyway, end of an era. See you later, Google.

How many collectors of curiosities and antiquities, signs and wonders the likes of which no one has seen, are there—that will never share a fraction of them, that are never heard, that are never seen?

It seems to me that all the best of everything is constantly slipping past everyone. We’re blind. Instead, the whole thing is about the shopping and the commodities.

It’s 5:00 pm on a Sunday evening and the list of “things that I still have to do this weekend” still runs into the dozens and any of them is hours long.

The truth is that my life these days is mostly neglected. Everything that is mine, me, about me, about my life, from friends to future to housekeeping to dental appointments is essentially forgotten. Neglected.

What have I been doing all weekend? Working, of course. Like a lot of Americans. We know this. It’s our culture. Every other movie and television program ever made is an implicit or not-so-implicit warning about living exactly this life, precisely because everyone does it. Do I blame my employer? Not really, no. I’m grateful to have work at a good job with a good title. Many, many people in tech right now don’t. I am getting paid to work on cutting-edge stuff. How could you be upset about that?

And yet it’s also clear that our society is broken. This is broken. I should not be at work on a Sunday evening, having worked all weekend long, wondering when I will get a chance to pop over to the drugstore, still unshowered, for just a few basics (maybe at 11:00 pm or midnight; it’s a 24-hour drugstore) so that I can start the week with toilet paper and soap in the house. I will work until I just can’t any more, as I’ve done all weekend, and then I’ll wake up Monday morning to the work week.

When will I mow the lawn? Engage in hobbies? Fix the plumbing? Go out and see the world?

Don’t be silly, that’s what rich people get to do.

Will I ever get to talk to a friend or do martial arts again? Who knows.

I need to figure out something basic about life. The problem with all the movies and TV shows is that they just contain the warning, not the solution.

All the people I’ve been in love with over the years, and none of them ever in love with me except when they were, in secret, beyond the reaches of empiricism or admission. No love possible in girlfriends, not in wives, not in muses in drunken stupors in houses of ill-repute. That’s not us. That’s not what we do, or could ever do.

The world is full of this music that nobody can hear, swirling and caressing everyone as the time flies past, past, past, and we dance, moths, around capital’s flames.

I don’t know what to say to any of you other than that I loved you and I love you still, or maybe I love the idea of you, the past of you, or maybe even the you that you never admitted to, that you wanted to disown, that you swore didn’t exist, but that I saw. That I saw.

I’m getting older. Time is passing. I don’t know what the rest of my time looks like. I know that I loved all of you, even if you hated my guts, or think you hate my guts. And that I’ll still be here for you when you call again, even though you swear you wont, ever, until you do.

We live in a broken time. I am probably broken. You are broken. I forgive us all for being broken. In another time, we would have been happy. But in our time, in the history that we belong to, this is not open to us.

But I have loved you. You know who you are.

As I sit here working, aging in place at 10:30 corporate PM:

is that men can’t ever both (a) tell the truth and (b) be loved (or even liked).

This makes life suck, but it’s our lot in life. Anyway, life mostly sucks… anyway.

The state of AI:

• The $180 of memory I bought when I built the machine I’m now using for inference would now cost $700

• The <$1,000 of GPUs (4x32GB) I bought in Q1 of this year would now cost $4,500

• But as a result everyone is moving to models that don’t need so much compute or bandwidth, e.g. Qwen 3.8 Flash Next with its n-gram sidecar

• Almost nothing is supported and nothing can keep up with the change

• All the AI stuff is so buggy that it barely works

• But you can have your AI empirically profile your hardware and then write or port drivers, applications, and inference engines (vLLM with all custom kernels took maybe 5 hours)

• And you can just pull the Github repo that’s buggy and tell your AI to fix all the bugs and submit PRs and then the application is perfect

• AI has “enabled” us and many other orgs to let a lot of people go who may never work in their field again

• But we’re desperate to hire more AI engineers and are actively giving young people, more conversant with AI, paid internships to do AI

• The AI world means that I sit here often for hours on end working with AIs

• But in the past I’d be sitting here for hours on end thinking alone in C or PHP or Python or JS, rather than chatting with another party in English

Is it the best of times or the worst of times?

It is the best of times. it is the worst of times.

— § —

In case you’re one of the many that acquired a happy family of Radeon Pro v620 hardware and now finds yourself wishing vLLM supported it, it does.

My related repos:

• vLLM for 2x/3x/4x Radeon Pro v620 (Qwen 3.8 27B and Qwen 3.5 122B builds), w/containers

• vLLM for 4x Radeon Pro v620 (Qwen 3.8 Flash Next build), w/containers

So… summer went quickly.

Kids have been back in school for two weeks already. And now I have two kids on Utah Valley University campus, which makes me feel old.

And wistful.

But whatever, life goes on.

— § —

And for better or for worse, right now a lot of “life” is AI and AI systems. At work, that’s my responsibility set in large part, which means that at home, that’s also what I do.

Current stuff:

• Just did some hackery to make vLLM work on RDNA/NAVI21 (i.e. Radon Pro v620s, as I’ve become increasingly weary of llama.cpp’s broken cache management. That repo is at: https://github.com/leapdragon/vllm-rdna2-recipe.

• Speaking of Radon Pro v620s, I’ve now got four of these. The last one still has to be mangled and bangled into the case, but it’s a mechanical problem at this point, we have a 1600w supply and risers to enable display and SAS out of other orifices (M2 slot, almost forgotten x1 slot).

• Now have both Openclaw and Hermes running, as well as a dedicated tool host that is its own metal with decent specs but oddly what I mostly seem to do is sit around generating code with Kilocode (and then getting teased for using Kilocode).

• Working on a reimplementation of a key favorite app from Newton OS. Yes, that Newton OS. There are parts of that OS that my brain is still shaped for and at this age I doubt I’ll ever recover.

And in general, it’s amazing to see Linux suddenly setting the world on fire. Apparently 2026 will be the year of Linux on the desktop, which tells us that I am exactly 33 years ahead of the curve.

— § —

I should post the tech stuff above on LinkedIn. “Hey I just ported vLLM to a new GPU” or “hey here are the things you need to know about agentive AI on local inference in 2026” or whatever.

But the thing is that a whole bunch of my sanity is maintained by not having to be a part of the corporate high school, keeping some semblance of dignity and integrity about me.

So, like a lot of other things I’ve done, it’ll just fade into the ether. I just can’t do it. I mean, I really hate LinkedIn. Like… really . hate.

But anyway, there’s a ton of code being written and hardware being hacked on. I haven’t been this personally embroiled in tech since the late ’80s and early ’90s. We’ll see where it goes.

— § —

In a way, the “where it goes” question is kind of important, because I’m soon going to be an “empty nester” and when I am, I’ll really be one. As in, empty.

Because I don’t live with anyone else. I don’t have a significant other. All of my best friends live out of state. It’s not clear to me what I’ll do, or why I would be motivated to do it.

I’ve passed that point in my life where work and ambition can be the things that get me up in the morning. So it’s got to be other things. But I’m not sure what.

— § —

Also amusing is that most of my interest in and attention to AI is really in a ’90s way, i.e. I’m still a Gen X computer guy at heart. That means:

• I really like banging on the hardware, things like trying to fit more and more into a case and trying to get exotic hardware to work and to integrate with other exotic hardware

• Though I ought to be sitting here building agents and harnesses, I have so far found that kind of boring

• What I am finding interesting is using AIs to generate code, not autonomously, but working together on app projects that have no real purpose any longer thanks to AI

I burn millions of tokens at times, but it’s not hundreds of millions and it’s not fully autonomous, because I want to be the architect; I want to manage what gets created, moment by moment.

In short, I’m stuck in the ’90s, but that’s sort of okay.

I even went and got a Cherry keyboard so that I have old fashioned somewhat-long-travel keys, and I’m sitting here with a trackball instead of a mouse.

— § —

Some model testing, subjective impressions:

Qwen 3.5 122B A10B: I really love this model. I wish it was newer, and had received the benefit of everything that’s been learned about training over the last half year. It’s code is perfectly fine, but it lacks taste and vision when it comes to tech work. With that said, it is a great reasoner, planner, and chatter, one of my favorites at any size, that knows a great deal. I’m not running it right now, but I have a big soft spot for it.

Qwen 3.6 35B A3B: If you run this at Q8, you’re getting a blazing fast model that writes great code and has great taste. It’ll one shot a lot of things. It’s not quite as good at getting into the real multi-turn nitty-gritty of hard-problem dev drudgery; at those times it can feel disengaged.

Qwen 3.8 27B: For coding and project tasks, this is the best local model I’ve ever tried. I thinks a lot. But it’s the first local model that gives you that “frontier feeling” when working on a codebase. If you try it, you’ll know what I mean.

Laguna S 2.1: Don’t bother. Qwen 3.8 reasons a lot, but then generates perfect output that surprises you with its quality. Laguna S 2.1 just generates output. A metric ton of thinking output that… goes nowhere. It will nervously and neurotically think itself in circles for many hours and if you force it to complete and generate, it will generate something that well, well behind frontier quality and also well behind Qwen 3.6 35B A3B.

Nemotron 3 Super: Nah. It’s not special and more importantly, it “feels” just like a bot. Ok for code but lacks the “strokes of genius” that you’ll see when coding with Qwen, and just doesn’t have much that’s interesting to say.

— § —

Somehow Saturday is almost over. The weekend just barely began and I’m already thinking about the work week.

— § —

What am I going to do with myself?