pseudocomposer

Are there specific problems? I’ve been primarily Ruby at work the last 6 or so years (before, also doing Ruby, but also lots of Java, Kotlin, Swift, FE stuff, and some Python)… exactly that period… and haven’t really noticed any issues. Granted, I don’t typically use a lot of obscure binary-only dependencies, or things whose C/C++/machine code wouldn’t mostly cross-compile fine.

Granted, all this time I’ve always only used rbenv-managed Ruby versions (and rvm before that). I’ve long disliked/avoided Python because its ecosystem lacked any equivalent, but I know “uv” has gained popularity there, and perhaps Mise is good in Python land as well.

What do you define as “class warfare?” Do you agree that the current status-quo hyper-consolidation of wealth our economy has fostered since act least 1972 is already an ongoing type of class warfare?

And finally, why do you think class warfare can’t get us anywhere?

I mostly just use LLMs for autocomplete (not chat), but wouldn’t this be fixed by adding a “delete message” button/context option in LLM chat UIs?

If you delete the last message from the LLM (so now, you sent the last message), it would then generate a new response. (This would be particularly useful with high-temperature/more “randomly” configured LLMs.)

If you delete any other message, it just updates the LLM context for any future responses it sends (the real problem at hand, context cleanup).

I think seeing it work this way would also really help end users who think LLMs are “intelligent” to better understand that it’s just a big, complex autocomplete (and that’s still very useful).

Maybe this is standard already, or used in some LLM UI? If not, consider this comment as putting it in the public domain.

Now that I’m thinking about it, it seems like it might be practical to use “sub-contextual LLMs” to manage the context of your main LLM chat. Basically, if an LLM response in your chat/context is very long, you could ask the “sub-contextual LLM” to shorten/summarize that response, thus trimming down/cleaning the context for your overall conversation. (Also, more simply, an “edit message” button could do the same, just with you, the human, editing the context instead of an LLM…)

I formerly worked at https://corista.com, dealing specifically with integrations with AI pathology algorithm vendors like https://www.qritive.com and https://visiopharm.com. (Note: we dealt specifically with microscopy rather than radiology.)

I really think we were doing things the “right way” before I left: providing AI analyses from various vendors as overlays for slides, and being able to pre-flag slides with obvious cases of cancer or other infection. (These analyses typically being custom algorithms provided by vendors through APIs we collaborate on, not LLM output.)

With things like this simply built into an existing, established pathology platform, I’m pretty confident a team of 4-5 pathologists could do the work of 6-8, with better quality output, similar to how Copilot and other tooling speeds us up as developers. At $300-400k/yr/doctor, that’s considerable savings (and the opportunity to allocate more doctors in specialties that aren’t easily automated).

However, for lots of reasons, it seems the market doesn’t necessarily agree with me on the value potential of this approach in this field (which can certainly be a self-fulfilling prophecy).

In the last few years I've definitely learned I get some sort of "release" from practicing/holding handstands and other hand/arm-balancing moves and, later, (less so, but with less weird looks from people) from yoga balances (tree, warrior 3, eagle, [revolved] half moon, dancer).

I agree in ways, but I think a single global fund is a bit far-reaching and over-centralized, thus prone to corruption. I think it would make sense for, say, the US, EU, and BRICS (with maybe China as a separate entity) to each have open-source funds for OSS digital infrastructure, and cooperate on global web standards. So if one fund goes rogue/corrupt or is crippled by Republican-types, the world still has two (or three) backups.

Those who fall between the two are usually those who didn’t have a family nest egg to start with (ie, their parents were factory workers rather than accountants).

Couldn't you just do this with a regular for loop and a few datatypes/functions? (This pseudocode is an Elm/Haskell/Rust/TypeScript-inspired abomination, but pretty portable to any language...)

    type Node = { value: Any, left: Node, right: Node }
    type Direction = Left | Right
    type TreePosition = { root: Node, currentNode: Node = root, position: Direction[] = [] }

    # Implementation left as an exercise but should be obvious and run in O(1), I believe. Returns Nothing when we're out of nodes.
    function nextPosition(position: TreePosition): Option<TreePosition>

    # The tree you want to iterate through
    const myTree: Node = ...

    # The loop
    for(let position: TreePosition? = TreePosition(root: myTree); position != Nothing; position = nextPosition(position) {
        node = position!.currentNode
        # Your loop code
    }

I'd argue this doesn't belong as a language-level feature, but maybe an API/stdlib-level feature.

Are you saying this includes the price of not being allowed to criticize the CEO?

I largely agree, but I also think you might want to consider how the existence of LLMs affects our education system. I think this is one of the places they really have the most potential to cause harm, but also perhaps some incredible (humanity-changing) good.

I suspect countries, and even individual local schools, who effectively adapt their curricula to account for LLMs will see an enormous difference in student outcomes in the next couple of decades.

If you can self-host or publish a demo of your work that is open-sourced, I find that that really opens the door for a lot of web dev positions, regardless of what particular languages you know and/or what’s in vogue.

(Mine would be https://jonline.io/jon, part of a larger project that does federated social media, but just a “hello world” portfolio server app that you open source would be quite effective, I think. This is all based on recruiters’, hiring managers’, and other devs’ responses when I link them to it.)

At least on iOS, I've found ScreenZen (https://www.screenzen.co) to be pretty ideal. I'll definitely try SpeedBump if they add me to the beta, though.

Like SpeedBump, ScreenZen also has a "wait before using addictive apps" function (though, it doesn't have the "wait a bit longer to get 15mins vs. 2mins" - that would be really nice).

ScreenZen also keeps track of "streaks" of days not going over your overall limit in addictive apps. Unfortunately, that limit is counted by number of sessions, not the actual overall app screen time (by default, if you set your limit to 1 hour, it gives you 6 sessions of 10 minutes per app). It would be nice to combine this with SpeedBump's different-time-limit feature: if I want to quickly check a post or show a friend something on Instagram, I could use only 2mins out of my overall hour limit, instead of 10mins.

I just got a RAV4 Prime (a “plugin hybrid” or PHEV).

In EV mode, it has 40 or so miles of range. That’s high enough that I never need to use gas during the week, unless I go to the next city over. It’s also low enough that it can always fully charge overnight without a special circuit.

If I need to go further, it just becomes a RAV4 Hybrid (with better performance - aka HV mode). No long waits or searching for chargers needed.

Generally I put in maybe a half tank of gas ($15) once a month. Especially in the US, I really think PHEVs make much more sense than EVs. BYD seems to make some really nice ones, but we’re sadly hostile to them.

I’m also a DigitalOcean user, but I prefer managed K8s and don’t think there will ever be a reason to go back to having to deal with host OS things again. I’d rather just pay for my CPU/RAM, and give it Docker images to run, than worry about all that. And DOKS (DigitalOcean K8s) doesn’t cost any more than bare DigitalOcean boxes.

Cert-Manager is a CertBot-compatible K8s service that “just works” with deployed services. Nginx ingresses are a pretty standard thing there too. Monitoring is built-in. And with a few API keys, it’s easy to do things like deploy from GitHub actions when you push a commit to main, after running tests.

And perhaps most importantly, managed Kubernetes services let you attach storage for DB and clusters with standard K8s APIs (the only thing provider-/DigitalOcean-specific is the names of the storage service tiers). Also the same price as standard DigitalOcean storage with all their standard backups… but again, easier to set up, and standardized so that if DigitalOcean ever gets predatory, it’s easy enough to migrate to any of a dozen other managed K8s services.

Various ways in various companies, but in my last few companies pretty much always “CI runs tests and builds/uploads a Docker image, CD deploys it onto a K8s cluster somewhere.”

In the case of my major current personal project, I do it with a GitHub Actions workflow: https://github.com/JonLatane/jonline/blob/main/.github/workf...

A deploy looks like this: https://github.com/JonLatane/jonline/actions/runs/1346474905...

Here, I do a canary deploy to a dev server (jonline.io), then cut a GitHub release and deploy to production servers (bullcity.social, oakcity.social). (All really live on the same single-box dinky K8s cluster.)

And then there’s https://en.m.wikipedia.org/wiki/Carcinisation

When you were remote, did you ever work from coffee shops or coworking spaces? In your hybrid role, are you actively working with others in-person?

I’ve definitely found that having other people around during some of my work week has a positive psychological impact for me, and solve it this way. At the coffee shops and spaces I frequent, I often see and cowork with folks I know from my various hobbies and clubs in town.

I remember that, even when I was in-office full-time (pre-2020), sitting in the same cube/desk area as my teammates… we mostly communicated over Slack about work. Only very occasionally would we come look at one another’s screens - less than I screen share over Zoom or Slack Huddle with them these days. (Slack Huddles are especially more efficient than anything that could be done in-person, as you can both view one another’s screens.)

I definitely did enjoy hanging with my colleagues in those days - really, socialization was the majority of our in-person interaction (well, and meetings, which I might argue are better conducted over Zoom, as some people being physically imposing, reducing helpful input from others, was always a thing).

I’d also say all our in-person hanging made us more of a monoculture - half my team would hit the gym downstairs and be spotting each other on bench press during lunch breaks. Whereas in my current remote role, I work with folks who never go to the gym, but are into dance, cooking, theater - all sorts of more varied things! I think the variety of life perspectives may make us more productive as well, though obviously there are too many variables at play to really compare these work scenarios with any sort of objective metric.

It seems like something like medical/legal professionals’ annual/otherwise periodic credential exams might make sense in fields where AI is very usable.

Basically, we might need to standardize 10-20% of work time being used to “keep up” automatable skills that once took up 80+% of work time in fields where AI-based automation is making things more efficient.

This could even be done within automation platforms themselves, and sold to their customers as an additional feature. I suspect/hope that most employers do not want to see these automatable skills atrophy in their employees, for the sake of long-term efficiency, even if that means a small reduction in short-term efficiency gains from automation.

The thing about being “poor” in Europe is, you can be “poor” your entire life by American standards of income/assets, but still never take on medical debt, travel more than a month out of the year, have a couple kids and educate them in good schools, and retire.

Most Americans cannot.

> What should they be doing once all the meaningful jobs are automated away?

My point is that looking at patient's biopsy slides or invoices, for 40-80 hours a week, to find known cancers/viruses, or to make sure numbers match up with statements, is not meaningful work. In fact, I'd argue that any work that can be automated without performance degradation isn't meaningful work for humans.

For the case of pathologists, this frees them up to look more carefully at cases where patients are experiencing symptoms and the cause can't be automatically. That means better patient care for both common and exceptional cases. And, perhaps, when put into use globally, it frees some pathologists up to take other roles in medical or scientific fields, doing something that isn't easily automated.

For accountants, it frees them up to only look at cases where things mismatch, or, well... science, teaching, art, local farming, carpentry, really anything that isn't staring at invoices and statements 40 hours a week. It seems particularly silly to me that anyone would argue "comparing numbers on documents" is "meaningful work."

A pretty popular quote by Stephen Jay Gould: "I am, somehow, less interested in the weight and convolutions of Einstein’s brain than in the near certainty that people of equal talent have lived and died in cotton fields and sweatshops." I'd assert that many of our world's pathologists and accountants are doing the equivalent of living and dying in those cotton fields and sweatshops, if they're doing work that could be automated.

> Stripping people of their livelihoods is one way to encourage them to overthrow the system, yes. Make no mistake though: the revolution will not be kind to you.

Chill, dude. It's not like I'm making millions building these things. It's all just SaaS software, a sector where profit margins generally always trend downwards as the market for the product matures. Put another way, these companies are nowhere near as profitable as companies that extract our psychological profiles, target content and ads to make us feel miserable, and then use that to sell us stuff (Meta, Google, et al). I could easily go to Meta and make twice what I do now.

Plus, I contribute time and labor to organizations like the DSA, JVP, and local nonprofits, volunteer at my local food bank and for other local organizations, choose to shop at local co-ops and farmer's markets, and more to build the type of economy that can support actual meaningful, not-easily-automated work for humans to do.

My last two roles have involved automating detection of cancers and other disease in biopsies (eliminating pathologists’ jobs in hospitals), and now automating invoice payment and statement settlement in the global freight sector (eliminating accountants’/operators’ jobs in these companies).

I’m honestly very excited about this type of work. The benefits to society - better healthcare for less money, humans with their beautiful, creative brains not spending 40 hours a week looking at documents unnecessarily - are incredibly obvious. The flipside of this progress is, of course, that we have to overthrow the extractive capitalist system we have in place. We’re able to do this work in spite of that wasteful, inefficient system, not because of it.

I think the only thing that will actually push people to overthrow it is exactly the type of work I’m doing. Much like the first industrial revolution begat our first efforts at creating a more socialist, just, and democratic society (eg, like Fiorello La Guardia for instance). The benefits to this the first time were extremely tangible.

Or there’s Rust, where types are very strictly mere containers for data and nothing else (until you add traits).

None take more than 5s (most 2-3s) on my iPhone 16 Pro on a 400Gbps fiber connection.

I built Jonline for this purpose. I’ve now deprecated the Flutter frontend, making it a Rust BE/gRPC API/React/Redux monolith. My blog/portfolio site is at https://jonline.io.

They are certainly diverging, and becoming more useful tools, but the answer to “which one can actually become AGI?” is, as always, “none of them.”

Well, Apple employees can’t work remotely :) But if you do, and you like working outside your house, the cheapest 10+-inch M-series model you can get is the answer. With Sidecar, it’s at minimum a wireless external display everywhere you go.

At this new model’s cost, it’s hard to recommend. But it does make last year’s M2 models (and older M1 models) more sensible for this purpose.

We’ve had it all along: small businesses and city/county governments.

What we’re missing is a willingness/ability to break up overreaching corporations and states. We used to be better about it - see 19th-century Europe, especially France - but fighting back obviously has costs.

Centralized power isn’t great at supporting human life, but it’s very good at dominating human life. So large corporations and states have been bedfellows and dominated us for the last century or so by greatly elevating the cost of actually fighting back against their united power.

Emotions are like waves; we can’t choose which ones appear, but we can choose which one to surf. The person you’re replying to (well, really, the person they quoted) didn’t really seem to be “attacking” anyone’s emotions to me. It seemed pretty gently advising the reader not to spend all their time riding the emotional waves that lead to depression or nihilism.

I do believe we all must face these emotions to become aware. But the depression/nihilism trap is very real for many people (myself included), and learning to walk that line and stay curious is part of our emotional/psychological development.

I wrote the comment so long mostly to appease people like you. But please, post unuseful responses and I’ll just assume this was the DOGE actually doing something efficient

I really liked the simplicity of his explanation in information theory terms. Thank you!