post-it

> Look at what it costs to get a work shirt (I mean, for physical labor, "blue collar", heavy chambray or something along those lines) of comparable quality & materials to what was in a Sears catalog in the 1930s or ordered by the US military in the 1940s, which in neither case could be regarded as super-fancy. You're probably looking at minimum $150.

Of course, this is still cheaper than it was in the 1940s. With my disposable income I could afford to buy a few $150 shirts a month. A worker of my social class in the 1940s could not.

I don't need the quality so I buy $5 Gildan shirts instead. I do buy Made in Canada cat toys for my little guy though. Different priorities.

I guess there's something to be said for "hey, if you're considering buying a Tesla, you may be the kind of person that's likely to kill themself in a car crash. Consider buying a safer car or taking the bus!"

For a while they were the safest car in crash tests, weren't they? Was there an inflection point where they were dropping like a rock? Or is this a case of measuring different things (crash tests vs fatal accident rates)?

I know you probably don't know off the top of your head, I'm hoping someone can chime in.

Usually when people provide examples, they're intended to serve as a representative sample of a larger trend, and not an exhaustive list. Hope that helps.

Thanks! Could it conceivably load the sub-models in series rather than parallel? 8 still won't be enough but I wonder if those with 16 could eke something out.

How much RAM does this use? Only sitting on 8 GB right now, I'm trying to figure out if I should buy 24 GB when it's time for a replacement or spring for 32.

It's not in the interests of the UAE to improve. There's the (possibly misattributed? but topical nonetheless) quote by the previous emir of Dubai:

> My grandfather rode a camel, my father rode a camel, I drive a Mercedes, my son drives a Land Rover, his son will drive a Land Rover, but his son will ride a camel.

They want to prolong the Land Rover phase as long as possible.

Well ultimately the skill I care about is understanding software, changing it, and making more of it. And clearly that isn't atrophying.

My syntax writing skills may well be atrophying, but I'll just do a leetcode by hand once in a while.

Do folks see a mention of Claude as indicative of an ad? I usually say it because at work we've got a couple different options and I like to mention which one in particular I was using. But maybe I'll just start saying AI on forums unless someone asks me to specify.

> however imagine that the new tool suddenly gets ragged pulled from under your feet

When you have a headache, do you avoid taking ibuprofen because one day it may not be available anymore? Two hundred years ago, if you gave someone ibuprofen and told them it was the solution for 99% of the cases where they felt some kind of pain, they might be suspicious. Surely that's too good to be true.

But it's not. Ibuprofen really is a free lunch, and so is AI. It's weird to experience, but these kinds of technologies come around pretty often, they just become ubiquitous so quickly that we forget how we got by without them.

> If you were on a plane flight with no wifi, and you wanted to do some dev work locally on your laptop, how comfortable would you be vs if you had done all that work yourself rather than via Claude?

Probably about as comfortable as I would be if I also didn't have my laptop and instead had to sketch out the codebase in a notebook. There's no sense preparing for a scenario where AI isn't available - local models are progressing so quickly that some kind of AI is always going to be available.

Yeah, I test everything myself.

For example, Claude was very eager to include function names, implementation details, and the exact variables that are passed between services. But all the info I need for a particular process is the names of the services involved, the files involved, and a one-sentence summary of what happens. If I want to know more, I can tell Claude to read the doc and find out more with a single query (or I can just check for myself).

I was worried about skill atrophy. I recently started a new job, and from day 1 I've been using Claude. 90+% of the code I've written has been with Claude. One of the earlier tickets I was given was to update the documentation for one of our pipelines. I used Claude entirely, starting with having it generate a very long and thorough document, then opening up new contexts and getting it to fact check until it stopped finding issues, and then having it cut out anything that was granular/one query away. And then I read what it had produced.

It was an experiment to see if I could enter a mature codebase I had zero knowledge of, look at it entirely through an AI, and come to understand it.

And it worked! Even though I've only worked on the codebase through Claude, whenever I pick up a ticket nowadays I know what file I'll be editing and how it relates to the rest of the code. If anything, I have a significantly better understanding of the codebase than I would without AI at this point in my onboarding.

> and deploy to hardware you own or rent.

Because this part sucks. I grew up fiddling with Linux. I don't want to play devops anymore. I want to write code and run it.

- You know, Lindsay, as a software engineering consultant, I have advised a number of companies to explore closing their source, where the codebase remains largely unchanged but secure through obscurity.

- Well, did it work for those companies?

- No, it never does. I mean, these companies somehow delude themselves into thinking it might, but... but it might work for us.

Not many existing models will do it or do it well.

> when you can do the fine tuning directly on the phone

Well you'd need millions of on/off nude samples to start with, so I don't think that's likely. And that's assuming we get a billionfold increase in mobile CPU performance or a billionfold decrease in fine tuning compute requirements.

You're reaching for things that, if they happen, are a decade or more out. It's ok to solve today's problems today. Perfect is the enemy of good.

> I edit it. I code around it.

You're never gonna guess what software engineers do.

Prohibiting apps from shipping with models that can nudify people seems reasonable. Custom models that can do it will be banned from search results and major sites. Just like child porn, they'll exist online but be quite difficult to find.

There is significant societal value in preventing crimes rather than just punishing them.

Nudify apps are very much present on the front page of search results. Get rid of that and you'll significantly cut down on the problem.

I doubt it. Photoshop was already pretty accessible, moreso than downloading sketchy models and getting them to run.

It's also a matter of scale. If a dozen teenage boys in a class are generating nudes, they feel safety in numbers. If it's only the weird computer kid who can do it, it's far easier to address and far more likely that it'll blow back on him.

> Like I said… skipping ahead, let me guess, For The Children; we must block local models, AI except for The NYSE Chosen Ones, encryption, and must have digital ID to use everything. Tell me this article is in favor of anything else.

No, delisting online nudify apps will take care of 99% of this. There's no reason for them to exist.

Teenagers aren't running generative models on their home PCs, they're using online tools. Sure, some people will do it offline, but these are by and large crimes of opportunity.

Yeah, we've pretty successfully gotten rid of online child porn from like front page search results and public social media, there's no reason why we couldn't do the same for this.

What the hell kind of queries are you running? I use Claude Pro all the time for asking questions, doing data analysis, writing side projects, and I very rarely get rate limited.

I use Claude Max 20x at work and I rarely hit 10% session utilization, which implies even using Claude to write code all day only uses 2x the Pro token limit.

Are you just telling it to try again when you get a response you don't like?

And they also make your WP admin page look like an IE6 toolbar.

I haven't tried it, but it looks like it just calls pbcopy, which won't work in a browser. There's no programmatic way to do it, because tmux can't access the clipboard.

Not in browsers, unfortunately. Both my work and Home Assistant OS use tmux in a browser tty, and I have to turn off mouse mode to copy.

It wouldn't help in this case, since the attacker was willing to pay.

Not for me. Some kind of OSC 52 problem.