apgwoz

16MB requirement! I wish all stuff was lightweight, like this! We’ve lost our way…

So sad. :(

Om was off my radar for the last 10 years or so, and then I recently encountered an article he wrote (https://om.co/2023/02/05/why-modern-leica-m-is-a-great-lands...) about his adoption of Leica M cameras. He had a wonderful eye: https://www.photosbyom.com/

How many layoffs does a company have to do before realizing it’s in their best interest to start asking other companies to take the employees they don’t want to employ anymore?

Also, 75% placement seems wildly successful. Why isn’t Cisco also a head hunting firm?!

It’s absolute baseline, but yes, it relies entirely on the platform’s permissions model, the administrator who assigns permissions, and the application authors to not create vectors for env var dumps. :)

But honestly, if you’re in the container, and the application running in the container can get secrets, so can a shell user.

_Maybe_ there’s a model where the platform exposes a Unix domain socket and checks the PID, user, group of the connection, and delivers secrets that way? This has its problems, too, like it being non-standard, only possible in some scenarios and otherwise fallible… but better than nothing? If you reap the container when that process dies, you can’t race for the same PID, at least. I dunno

The Ed Zitron rant will be phenomenal.

Ah. The article has since been updated to point out that it’s not plaintext, but encrypted at rest (which would be expected). OK.

You’re thinking too much. When you run the app, the system decrypts the secrets and makes them available as env vars (or some other mechanism).

In an admin ui, you list the names of secrets only, and provide a “reveal” or a “replace” on each one. They are never decrypted unless explicitly asked for.

Is this perfect? Absolutely not. The key is controlled by the company, but it can be derived in a manner that doesn’t allow for the dump of everything if it’s leaked.

Not at all. I was simply making a joke on the parallel exploration of states.

You pretty much have to assume someone is going to put sensitive data in an input like this. Encryption by default is the only sensible choice.

Yeah, but… what if?

And his successor John Turnip.

Yes! I’ve been trying (and failing!) to get people to understand this. Build the high leverage tools while the tokens are cheap. Unfortunately, I haven’t figured out the right set of high leverage tools. :)

As another data point, I pay for Pro for a personal account, and use no skills, do nothing fancy, use the default settings, and am out of tokens, with one terminal, after an hour. This is typically working on a < 5,000 line code base, sometimes in C, sometimes in Go. Not doing incredibly complicated things.

It takes longer, but a spade is better than bare hands. The goal is to speed up finding valid vulnerabilities, and be faster than humans can do it.

I use the models to look for vulnerabilities all the time. I find stuff often. Have I tried to do build a new harness, or develop more sophisticated techniques? No. I suspect there are some spending lots of tokens developing more sophisticated strategies, in the same way software engineers are seeking magical one-shot harnesses.

Why? They claim this small model found a bug given some context. I assume the context wasn’t “hey! There’s a very specific type of bug sitting in this function when certain conditions are met.”

We keep assuming that the models need to get bigger and better, and the reality is we’ve not exhausted the ways in which to use the smaller models. It’s like the Playstation 2 games that came out 10 years later. Well now all the tricks were found, and everything improved.

The benefit here is reducing the time to find vulnerabilities; faster than humans, right? So if you can rig a harness for each function in the system, by first finding where it’s used, its expected input, etc, and doing that for all functions, does it discover vulnerabilities faster than humans?

Doesn’t matter that they isolated one thing. It matters that the context they provided was discoverable by the model.

There’s no doubt that stuff is print making. My point is that there are multiple ways of doing (within each of these): relief, Intaglio, lithography, screen printing, offset.

So if you say, “I’m a print maker,” it describes basically nothing. :)

This is just a general statement, not directed at you. Sorry it felt that way.

I like your stuff! I’ve been coveting a plotter for a while, but I’m pretty sure it won’t get used enough to justify the expense. :/

I do find the term “printmaking” hilarious because there’s just sooo many ways to make prints. I tried to get into linocut fairly recently, but the battleship grey linoleum I had wasn’t very good. It cracked and crumbled pretty easily. I did get some of pink Speedball “blocks,” but it gets expensive pretty quickly. I guess more to the point is the feeling that I lack much to say. But, that’s an excuse. :)

You _can_ do trampolines, but that is kind of infectious, or needs to be very explicit with extra code, etc.

It’s also deterministic, unlike llms…

That's like a truck company using horses to transport parts. Weird choice.

Easy way to claim more “horse power.”

https://sigusr2.net/lemons.html

The point this article makes, that suddenly agents can do the work of customizing free software, completely makes sense. But, the reality is that the Free Software movement is opposed to the way Lemons are built today, and would not accept a world like this. (Rightfully!)

My belief is that Lemons effectively kill open source in the long run, and generally speaking, people forget that Free Software is even a thing. The reasoning for that is simple: it’s too easy to produce a “clean” derivative with just the parts you need. Lemons do much better with a fully Lemoned codebase than they do with a hybrid. Incentives to “rewrite” also free people from “licensing burdens” while the law is fuzzy.

The key to this argument is that we won’t need to rely on Anthropic/OpenAI soon — will they exist in the same way they do today in 12-18 months? The “open” models are getting better and better, and people are figuring out ways to make inference run on lesser hardware. It already might be viable for people that don’t expect “instantaneous” and are doing more hybrid development.

But you’re also never going to convince the people who still only run vi on the Linux console, without Xorg…

I was hoping someone made this comment! It remains high on my list of Frontalot songs. Big fan of “I’ll Form the Head” and “Stoop Sale” also from that album as well.

I honestly don’t remember what the frame rate was, but it definitely improved when I upgraded to a Pentium 100. I distinctly remember a buddy giving me some RAM (2x4MB) which allowed me to play on the 486. I was so happy!

The DX2s _were_ a significant improvement over the 486DX, but I’ll admit, I might be remembering the excitement of getting to play Quake at all! The framerate may have been 15-20 fps and I just dealt with it,

The minimum requirements, on the box, were apparently Pentium 75Mhz. 8MB (DOS), or 16 RAM (WIN95).

I played Quake on a 486 66Mhz DX2 with 16MB of RAM in the 90s. On the lowest resolution, but it was fine.

Whoops! Not spent, raised.

Anthropic claiming that its total revenue since January 2025 as $5 billion contradict that its expected run-rate revenue for the year 2026 is $19 billion?

Isn’t the “exceeding $5BN” comment a lifetime revenue? … on $30BN (edit: previously said spent) raised (or something ridiculous.)

A lot of the commentary on the frontier model companies is based on how much money they’ve spent to the relatively small amount they’ve made in return, and the skepticism, especially given almost continuous reporting, that deploying AI in a variety of situations doesn’t seem to yield favorable business outcomes. OpenAI shifting to enterprise / coding type stuff this week seems, also, potentially informative. Is Gen AI actually useful for anything but code? Signs keep pointing to no… and even then, we’re in the early stages of figuring out how to build without destroying everything… something Amazon just recognized as possible with their recent shopping outage.