jscd

I don't think the "would not exist if I hadn't created it" argument really works. If I delegate a project to someone else, and they do the project, does that mean I did it? In my mind, I could comfortably say I managed it, or I designed some parts of it, but I really couldn't take the credit for actually doing it. You wanted it to be done, so you asked someone to do it.

If anything, I would say your conclusion should be flipped: you didn't really do the "making," but you can still be proud of what you _did_ do: provide direction and a vision for a solution to a problem.

Hot tip: if you disable JS on the site, then nothing but the title loads and you can disregard it without wasting all those CPU cycles.

Does that change anything? We're still relying on OpenAI's account of where the LLM was running, what sandboxing restrictions were in place, the task it was given, etc.

Even assuming they're telling the truth about what this LLM's goal was, they still have motivation to be less than honest about the state of their "highly isolated environment." Either this model was really operating in a truly locked down intranet and it really did a series of highly complex lateral movements and privilege escalations in order to escape it... Possible, but incredible.

_Or_, the "highly isolated environment" was less secure than they make it out to be, and now they have to choose between a) admitting they let these models with security precautions disabled run in YOLO mode, with the only significant precaution being a third-party proxy server, _and_ their security team didn't notice a huggingface blitz happening on their network during a weekend, all of which seems reckless and negligent; or b) lying about the state of their internal security, dodging accusations of irresponsibility, and now they get to also claim their product is so advanced they can't even contain it.

I mean yes, you never “owned” the IP, but this doesn’t change anything. You also never owned any movie on a DVD, song on a vinyl, or text in a book. Software is a strange beast in the world of copyright law, but, at least in my lay understanding, video game discs still have had protection under the first-sale doctrine, which means you were always allowed to resell, rent, give away, or destroy your copy. Ownership of the medium is still something.

Maybe not in the millions, but Meta is certainly not free from bloodshed. For example, in efforts to promote "engagement," they left the rollout of Facebook in Myanmar dangerously unmoderated, and (at least according to claims by Amnesty International[1]) are at least partially responsible for the genocide of the Rohingya there, which saw the tens of thousands of deaths.

[1] https://www.amnesty.org/en/documents/asa16/5933/2022/en/

First, I agree it's cool that Atari, with all its ability to completely screw small projects over, didn't do that in this case.

But, at the same time, I find it interesting that "emulations and clones" are considered entitlement (in a derogatory sense), but copyright protection is not. Before 1976 in the US, the _maximum_ copyright term was 56 years, and that would require filing for an extension from the default of _only 28 years_.

I think it's easy to forget that copyright as we know it is not set in stone. Historically, after 28 years, most works became public domain and that meant you could do literally whatever you want with it and it would not be legally stealing at all. I think we as a society have forgotten what it means to have a public domain.

Wow, impressively insufferable

I can kinda see your point, especially if the meaning is still obvious and the tone is inviting you to participate, but I think you're misunderstanding what "privilege" means here.

It isn't a superior "flexing their privilege" over their subordinates. The superior doesn't care. They don't even think about it. Because they have power over you, they can just speak gibberish and you have to figure it out. In my opinion, a good boss should have enough respect for me to not waste my time by forcing me to decipher a thought they didn't even read before sending.

Tailscale itself only uses sqlite[1], so I’m not sure if that really holds in this case.

[1] https://tailscale.com/blog/database-for-2022

Other way around. To quote the article:

To conduct the literature review, the sugar industry paid the Harvard scientists the equivalent of $50,000 in 2016 dollars [...]

So it was actually about ~$5,000 in 1965 dollars.

This is not providing the same functionality as a "traditional VPN," in the sense that it does not do anything to your traffic going to the wider internet. With popular VPN services, they are an encrypted tunnel for all your internet traffic (some use the same protocol, WireGuard), but at the end of the tunnel they decrypt the message and send it to whatever website you requested, which is exactly what can cause those privacy issues you describe.

In this case, though, it creates an encrypted tunnel _only between your own devices_. This allows you to connect to all your devices, home desktop, phone, laptop, as if they were on the same network, allowing you to do fairly sensitive things like remote desktop without having to expose your machine to the public internet or deal with firewall rules in the same way.

Assuming this project is legitimate, then the only traffic this service would even touch would be those between your own devices, nothing related to public internet requests. And, on top of that, the requests should be encrypted the entire way, inaccessible to any devices other than the ones sending and receiving the requests.

There are many caveats and asterisks I could add, but I think that's a fairly straightforward summary.

Yea this seems to work great. Let's take a look at the entry on the site itself:

https://encyclopedai.stavros.io/entries/encyclopedai-artific...

Early prototypes of EncyclopedAI emerged in 2007 when librarian Margaret Chen noticed that her pet parrot could predict which encyclopedia volumes patrons would request by observing their facial expressions. This observation led to the first algorithmic models, which attempted to replicate avian pattern-recognition through neural networks. The subsequent integration of natural language processing in 2011 marked the system’s transition from experimental prototype to operational deployment across major American public libraries.

The system’s backbone consists of distributed servers housed primarily in repurposed bowling alleys, which Chen discovered provided optimal acoustic conditions for server cooling. EncyclopedAI’s training dataset comprises approximately 2.3 billion encyclopedia entries, supplemented by 400 million hours of recorded reference desk conversations and—controversially—dreams reported by participating librarians.

This is honestly very disappointing. Not using LLMs, but the complete lack of transparency about their usage. You can already see in the repository issues related to hallucinations[^1]. This is _fine_, but not if you seem to obscure the fact that these can be very, very wrong. This seems to only be mentioned in the very brief loading screen and at the bottom of the about page[^2]. Also, apparently many of the "core RSS feeds" are just... reddit[^3]???

For me, this is only useful as a curated list of news feeds (and subreddits I guess), but nothing more.

[1] https://github.com/kagisearch/kite-public/issues/97#issuecom...

[2] https://kite.kagi.com/about

[3] https://github.com/kagisearch/kite-public/blob/main/core_fee...