caconym_

> perhaps only in English

Wouldn't it work better to just write the thing in whatever language they can actually write in and then do a straightforward translation in a single pass?

> someone who is a lousy writer with deep domain knowledge going back and forth with the LLM to express some insight or communicate some information that the LLM would not produce on its own

This sounds reasonable on its face, but how often does it actually come up that somebody can't clearly express an idea in writing on their own but can somehow get an LLM to clearly express it by writing a series of prompts to the LLM?

And, if it does come up, why don't they just have that conversation with me, instead?

What is the value of this "output"? If I want to know what LLMs think about something, I can go ask an LLM any question I want. For a comment on [a site like] HN, either the substantive content of the comment originated inside a human mind, or there is no substantive content that I couldn't reproduce by feeding the comment's context into an LLM. At the extreme, I don't have any interest in reading or participating in a conversation between a bunch of LLMs.

The Trump administration is 100% responsible for setting up the conditions where this kind of breach is effectively inevitable. They created "DOGE", staffed it with (among other specimens) teenage hackers with established records of malfeasance and names like "Big Balls"---presumably without any serious attempt at checking backgrounds and/or responding appropriately to any findings---and (by many accounts I've seen) granted them the authority to demand root level access to government systems without auditable logging or any other record of their actions. There appears to have been effectively zero oversight within "DOGE" itself, and the organization evidently failed to accomplish its stated goals by an enormous margin. AFAIK The Trump administration never publicly acknowledged any of this or took any visible steps to investigate the allegations.

If I was aware of any remotely comparable precedent in any recent administration, I would certainly criticize them for it. But the "DOGE" episode was so far beyond the pale that I can't think of anything else like it.

Yes, and they did so because of their specific circumstances and beliefs.

It's almost as if there's nothing special or unspecial about any of these populations. Just transient cultural factors that (in addition to generally being understood in limited hindsight and through rose-tinted glasses) will inevitably erode and dissolve under sustained attack.

> lay down and crawl along and accept everything without even a whimper.

People just want to live their lives. Maybe you think you would be doing differently in their position, but until you've had a chance to prove it, I don't believe you.

Isn't Lutnick literally the chief architect of Trump's tariff policy? I can hardly imagine anybody more responsible for creating this outcome besides Trump himself, who would presumably have appointed somebody else if Lutnick hadn't been available.

> I don't see how he had a better view of the whole thing than anyone else.

Given the above, you really don't think Lutnick had a "better view" of the likely outcomes and timelines, including the Trump admin's planned and gamed out responses to certain outcomes, than the average Joe on the street? I think that's extremely, uh, naive.

somewhat comforting to know that this kind of reflexive fan bitching about departures from canon has been around forever

So what? It's an argument that maybe we should do something, for the American people's consideration, but it doesn't change the fact that this war has been started without Congressional approval by a president who IIRC explicitly campaigned on "no new wars" or something to that effect. As far as I can tell, he has stated no consistent or coherent justification for any of this.

If we are going to engage in brute force regime change on the other side of the world, regardless of how bad the current regime is (yes, they are evil), this is not how it should happen. And everyone should take a moment to remind themselves why the term "regime change" carries the connotations it does.

At this point, do you really think any of these "labs" would give up competitive parity or advantage just because they're already making life worse for a lot of people and (by their own admission) stand to make it much, much, much worse? The persistence forecast says no.

There are maybe a few token exceptions, like Anthropic's current pushback against the DoD, but by and large I think we can continue to expect them to pay lip service to safety while continuing to build toward systems that, by their own admission, have incredible potential to cause harm. As you noted, the fact that they employ safety researchers does not necessarily mean that they will put safety over revenue.

If you classify Pete Hegseth as a person, then yes, apparently. Or perhaps he's only into the domestic surveillance angle---IIRC those are the two things Anthropic doesn't want anything to do with.

> if I read a thought-provoking, original, enriching comment on HN, do I really care if it was actually written by a dog?

I would rather just read the thought as it was originally expressed by a human somewhere in the AI's training data, rather than a version of it that's been laundered through AI and deployed according to the separate, hidden intent of the AI's operator.

Wow, thank you for this incredibly valuable perspective. I guess I should just be happy I can still own my computers!

Does this mean I will be able to buy RAM for the NAS build I left too late (stupid me assuming I could rely on some basic modicum of price stability)? I assume not.

I have the disks but only random old gaming PCs to put them in. I think I'm going to expand my Proxmox cluster and run Ceph so that I don't have to pay that 6x markup or whatever the fuck it is these days.

I'm tired, man. I'm tired of living in this world where AI is simultaneously an unstoppable eschatological juggernaut that's already making everything worse and at best is going to steal my livelihood and destroy my family's future, but also a hype driven shell game with full buy-in from world leaders and the moneyed elite who see a golden opportunity to extract unprecedented amounts of wealth for themselves before the West falls and they have to make other arrangements.

The fact that the federal government is unwilling to restrict guns and other real causes of ongoing public health crises (such as massive passenger cars and trucks) even as the deaths pile up does not mean that any level of government should be piling onerous regulations onto other things that demonstrably cause essentially zero harm at the macro scale, such as 3D printers, non-commercial/non-military UAVs, and so on.

If the number of people killing themselves with 3D printed guns is not literally zero or vanishingly small at most, I would be very surprised.

So what? I haven't named any names. You can imagine whoever you like having control of it. If you think that control will be wrestled from the execs in some kind of internal power struggle, that just puts new people in the position to make the same decisions.

I'm going to assume this is meant as a reply to https://news.ycombinator.com/item?id=47043601 but my answer is the same: why (in this hypothetical scenario) are you assuming the AI owners aren't members of all three of these classes?

> sociopathic trillionaire brains

> men with guns and a very thin technocratic class

Bold of you to assume these are disjoint. In the US we're spiraling toward oligarchy with a billionaire president whose primary goal in office is evidently to enrich himself and his family, and his regime is (with the $encouragement of AI titans) aggressively pushing integration of AI in various government and military applications.

If I was somebody like Sam Altman and I had indeed been thinking about this, I'm not sure things could be going much better for me wrt. that scenario.

"Class privilege" doesn't go far enough. If the class war is a war then AI may be its Manhattan project: a weapon that destroys the value of, and the ruling elite's need for, labor.

I continue to be shocked by the way people (with platforms) talk up to the (speculative) line where AI replaces most or all jobs, and then lamely suggest that this will be bad for the people who've lost their jobs because they will be poor, or something. No. What actually happens in that scenario is that money ceases to have value, at least in the way we currently understand it to. That scenario will produce a handful of monsters—sociopathic trillionaire brains encysted in layers of automation and automated production—that will crave more resources, more land, more power, and they will fight each other by various means for those things, and the rest of us will be at best in the way.

This scenario is not a given, because it's not obvious that AI can become this capable in the near term where the stage is set for such a profoundly lopsided outcome, but you can bet these people are thinking about it now, if not talking about it, if not materially preparing for it. And they are, indeed, the only people with reason to feel optimistic about it.

> So it's not actually an architectural issue

I mean, maybe? It's not obvious to me that Zulip is just as easy to develop a third party client for as Matrix.

> in practice

What's true in practice is all that matters to me, because I have a real life use case. The only might-bes I care about are the risks.

I've been using desktop Linux for 20 years in various capacities. I most recently picked Bazzite for my gaming/desktop VM because it's simple to set up and works better than other distros I've tried to game on in the recent past. It's that simple.

Thanks for the input. I'm working on setting up an instance, so we'll see how it goes!

> The federation of Matrix seems risky to me to the person self-hosting. I don’t want to host random people’s content. I’ve read some interesting articles about the design flaws of Matrix that led me to believe that it’s not a good option.

You can just not federate.

My experience with Matrix as a "discord replacement" at the scale of a few tens of people is that it works and is stable but is also jank and has a lot of features I don't really care about, federation being one of them. Hence my enthusiasm for a possible alternative.

> What is confusing to you about the community tier? It is basically describing any type of community of people who are not a for-profit business. Groups of friends, non-profits, volunteer groups, etc.

The part that isn't clear to me is how you prove to the Zulip people that you are worthy of being included in that tier. I'm certainly willing to write a few sentences explaining my use case, but the help page is light on specifics.

Nothing, but there already exist many other Matrix clients (shitty as they may be at present), as well as (IIUC) an Element PWA that uses web push (which is IIUC supported by Synapse) for notifications. Synapse also (IIUC) can be configured to use an arbitrary push gateway.

This is what I mean by "generic" in the other comment you replied to. I appreciate the value of tightly integrated server and client applications, and fully believe that Zulip's implementation of notifications may be both a) better for usability and b) a lower maintenance burden for the development team than supporting web push in a PWA, but---again---I am looking at this from a certain perspective where the way Matrix is architected and the breadth of the ecosystem imply less long term risk for my use case.

> Can someone explain to me why we should do engineering work to build features where the stated objective is to help corporations use our product without paying for it?

I'm not sure what you mean by this. I am looking for a replacement for Discord for my small community of friends, and before today I had never heard of Zulip and knew nothing about its pricing or policies or history.

Thanks for the response. I'll discuss it w/ my users.

> That said, I'd like to push back on the idea that charging businesses for a tool that's an important part of their daily work "breaks the seal". Organizations with a software budget should be happier to pay a fair price for ethical, user-first software from a friendly vendor than for a closed-source product from a megacorp. And Zulip's full-time development team should be able to make a living building ethical FOSS software.

I think you touched on the sort of thing I'm concerned about with your mention of enshittification, though I think you're probably right that VC funding is involved in most cases. It is good to know that you've been at it for a decade and have (apparently) built a sustainable business selling a product people like.

My concerns (which I hope are understandable) aside, I certainly support your right to charge money for what you've made, as I said here (https://news.ycombinator.com/item?id=46953048).

I'm sorry, would you rather I had framed this post as an aggressive critique of the Zulip developers without addressing my own context? I think anyone who has seriously tried to use Matrix as a chat app rather than a chat app but also an expression of one's principled preferences for federation, decentralization, and e2ee everywhere will know exactly what I'm talking about.

I don't mean to shit on Matrix either. It's a hard set of problems they set out to solve, and Matrix is usable and legitimately self-hostable.

Yeah. This is exactly my worry: as soon as solutions to technical problems like this start going in the direction of "we'll offer a monolithic solution and charge users for access to it" instead of "we'll make it as generic as possible even if the alternatives for now are flawed", it makes me wonder about the long term trajectory of the project.

I don't mean to cast aspersions on the developers—I respect everybody's right to try to get paid for good work, and this looks like good work. I am just not convinced it's the right option for my specific needs.

I understand that (IIUC in Matrix the client decides what push gateway to use, and the Element client just hardcodes matrix.org and lets anyone use it for free), but it doesn't really do much for my practical concerns. I'm looking for something my users can tolerate (which means no monthly fee) and that I can be reasonably confident won't rugpull us or vanish in the next ~10 years.

I'm asking because I hate Matrix and actually want you to convince me: why should I accept the risk of migrating my friend group from Discord to Zulip, which has already "broken the seal" of restricting features behind a monthly fee even for self-hosted users, when I could migrate us to Matrix instead? Matrix seems like the much less risky option.

I see that you have a "community" tier that's free and doesn't restrict notifications, but it's not clear to me exactly what's involved in proving that we should qualify.