“probably 1% of the companies will stand out and become huge and will create a lot of value, or will create tremendous value for the people, for society. And I think we are just going through this kind of process.”

Baidu is huge. Sounds like good news for Baidu!

Kernel.org, home of the Linux kernel, hasn’t changed much.

Kernel.org today:

https://kernel.org/

Kernel.org in 1998:

https://web.archive.org/web/19980130085039/https://kernel.org/

Not a website, but since you mention BBSes…one thing that would look pretty familiar to a 1990s Internet user would be most of the text-based MUDs, the ancestor of MMORPGs, that are around.

The MUD Connector is still around, and still has a list of active MUDs.

While I suspect that most dedicated MUDders use dedicated clients, the base protocol is still normally telnet, and you can use a plain old telnet client to play…a protocol that predates Internet Protocol itself.

I think that “best” is open to various interpretations.

The most-emotionally-impactful in the context of the game?

The most-graphically-impressive?

The best-integrated with the game?

I often don’t try and play the latest-and-greatest games, and while I’m sure that I’ve played games with thunderstorms in them, I can’t immediately recall any recent first-person 3D games…and I’ve kind of shifted way from FPSes in recent years. Probably the newest 3D game that I can immediately recall playing that I distinctly recall having thunderstorms – though I think that they were rain is modded Fallout 4; I was using one of the weather mods.

I think it was one of:

• Natural and Atmospheric Commonwealth

• True Storms

There are radstorms that impact gameplay by dosing the player with radiation, and I suppose could be considered to a different form of thunderstorm. These are separate from normal storms. Fallout 76 also has radstorms, but they are less-frequent and far-less-damaging than in (modded, don’t recall base game) Fallout 4.

I guess that that’d probably be the most-graphically-impressive that I personally can recall off-the-cuff. I’m sure that there must be some newer, fancier thunderstorms out there.

For impact…I can’t recall for certain whether-or-not there was actual thunder and lighting other than in cutscenes, though there’s certainly rain… But The Saboteur is an Assassin’s Creed-style game (I understand; I’ve never played more than a very small amount of those games) set in World War II Paris. The areas that are occupied by Nazi forces are mostly black and white, with a small amount of color, mostly red, and at least some of the time, it’s raining. The areas where forces have been pushed back look kind of like spring. I think that it added to the game’s atmosphere a lot.

My understanding – I’ve never used it – is that Bluesky uses some sort of “curated feed” list. The idea, from what I gathered, is that some person (or people?) could create a list of stuff and then people subscribe to it. Seemed like an interesting approach, since it’s a route to improve personalizing content relative to, say, Reddit. Originally, Reddit intended to run off a recommendation system, but that kind of fell by the wayside in the first few years.

I’ve wondered how practical it would be to have people publish feeds, then take into account one’s voting behavior and how it reflects feed content to help do recommendations. Can’t just score a feed by aligned posts – otherwise, it’d be trivially-gameable you could have people spamming by creating feeds and including popular things, and then also including some spam item. But I could imagine that being the foundation for something that does a good job of recommending stuff.

I mean, there’s plenty of anime pornography on the Threadiverse too. It’s just that sopuli.xyz, your home instance, isn’t federated with a number of hosts (and you may not be viewing its “all” feed).

https://sopuli.xyz/instances

Look at the “Blocked instances” tab. You’ve got stuff like:

https://lemmynsfw.com/

https://kbin.burggit.moe/ (which I can’t seem to reach due to some sort of TLS issue, but burggit.moe proper has “Free expression, including the Loli/Shota/Cub variety, are welcome here!”, and I assume that this is a gateway to the same material). I definitely remember that burggit.moe used to deal with consentual-nonconsentual material and underage anime material, because it caused lemmynsfw.com to defederate from them.

https://lolicon.rocks/

https://ac.akirin.xyz/ I don’t know what content they truck in, and their front page doesn’t indicate it, but it looks like the scrolling URLs in the bottom contain a bunch of links to various Fediverse hosts that deal in underage anime porn, and the user icons seem to all be anime girls, so I’m assuming that that might be their thing.

Not going to do a complete list of the blocked instances there, just pointing out that even if you look at your “all” feed on sopuli.xyz, it might not be representative of the Threadiverse as an aggregate.

I think that there’s a legitimate place for all-in-one “smartphone” SoC PCs. You can make them cheaper, smaller, and use less power.

It’s just not really what I want for myself in a PC. I want the modularity and third-parties competing to provide components.

But I am pretty sure that there are plenty of people who don’t care about that.

There has to be enough scale to support products like that, though. SoC systems might cannibalize enough to make scale hard.

sigh

Well, we’ll see where things go.

If this is you, then build your own home server.

While I don’t disagree, there’s also a very considerable cost difference here between running locally and remotely.

If a user sets up an AI chatbot, then has their compute card under average 24/7 load of 1% – which would require averaging, say, a daily session for an hour with the thing averaging 25% of its compute capacity during that session – then the hardware costs for a local setup would be 100x that of a remote setup that spreads load evenly across users.

That is, if someone can find a commercial service that they can trust not to log the contents, the economics definitely permit room for that service to cost less.

That becomes particularly significant if one wants to run a model that requires a substantial amount of on-card memory. I haven’t been following closely, but it looks like the compute card vendors intend to use amount of memory on-card to price discriminate between the “commercial AI” and “consumer gaming” market. That permits charging a relatively large amount for a relatively small amount of additional memory on-card.

So an Nvidia H100 with 80GB onboard runs about (checks) $30k, and a consumer Geforce 4090 with 24GB is about $2k.

An AMD MI300 with 128GB onboard runs about (checks) $20k, and a consumer Radeon XT 7900 XTX with 24GB is about $1k.

That is, at current hardware pricing, the economics make a lot of sense to time-share the hardware across multiple users.

Oh so both hashes and synmetric cryptography are secure entirely by doubling up the key size.

That’s not my understanding, which is that it’s more-secure than that and doesn’t require the doubling. Assuming the pages I linked are correct and that the understanding of them from my skim is correct, both of which may not be true:

• About a decade-and-a-half ago, it was believed that AES of existing key lengths could be attacked via a known quantum algorithm – Grover’s algorithm – using future quantum computers. However, the weakness induced was not sufficient to render AES of all key lengths practically vulnerable. it would be viable to simply increase key lengths, not redesign AES, sufficient to make it not attackable via any kind of near-future quantum computers.

• At some point subsequent to that, it was determined that this attack would not be practical, even with the advance of quantum computers. So as things stand, we should be able to continue using AES with current keylengths without any kind of near-future quantum computer posing a practical risk.

Take all that with a huge grain of salt, as I’m certainly not well-versed in the state of quantum cryptography, and I’m just summarizing a few webpages which themselves may be wrong. But if it’s correct, you were right originally that there aren’t going to be near-term practical attacks on AES from the advance of quantum computing, not from any presently-known algorithm, at least.

So, I haven’t read up on this quantum attack stuff, and I don’t know what Kairos is referring to, but setting aside quantum computing for the moment, breaking a cryptographic hash would simply require being able to find a hash collision, finding another input to a hash function that generates the same hash. It wouldn’t require being able to reconstitute the original input that produced the hash. That collision-finding can be done – given infinite conventional computational capacity, at any rate – simply from the hash; you don’t need additional information.

I’m not sure I follow. Could you expand on that?

EDIT: Wikipedia says this:

https://en.wikipedia.org/wiki/Post-quantum_cryptography

In contrast to the threat quantum computing poses to current public-key algorithms, most current symmetric cryptographic algorithms and hash functions are considered to be relatively secure against attacks by quantum computers.[2][11] While the quantum Grover’s algorithm does speed up attacks against symmetric ciphers, doubling the key size can effectively block these attacks.[12] Thus post-quantum symmetric cryptography does not need to differ significantly from current symmetric cryptography.

The citation there is from a 2010 paper, which is old and is just saying that this is believed to be the case.

This page, a year old, says that it is believed that the weakening from use of Grover’s algorithm is not sufficient to make AES-128 practically breakable, and that at some point in recent years it was determined that the doubling was not necessary.

https://crypto.stackexchange.com/questions/102671/is-aes-128-quantum-safe

Keeping in mind that I am about twenty years behind the current situation and am just skimming this, it sounds like the situation is that one cannot use an attack that previously had been believed to be a route to break some shorter key length AES using quantum computing, so as things stand today, we don’t know of a practical route to defeat current-keylength AES using any known quantum computing algorithm, even as quantum computers grow in capability.

Because AES is NOT vulnerable to quantum computing.

I have not been following the quantum computing attacks on cryptography, so I’m not current here at all.

I can believe that current AES in general use cannot be broken by existing quantum computers.

But if what you’re saying is that AES cannot be broken by quantum computing at all, that doesn’t seem to be what various pages out there say.

https://crypto.stackexchange.com/questions/6712/is-aes-256-a-post-quantum-secure-cipher-or-not

Is AES-256 a post-quantum secure cipher or not?

The best known theoretical attack is Grover’s quantum search algorithm. As you pointed out, this allows us to search an unsorted database of n entries in n−−√ operations. As such, AES-256 is secure for a medium-term against a quantum attack, however, AES-128 can be broken, and AES-192 isn’t looking that good.

With the advances in computational power (doubling every 18 months), and the development of quantum computers, no set keysize is safe indefinitely. The use of Grover is just one of the gigantic leaps.

I would still class AES as quantum resistant, so long as the best-known attack is still some form of an exhaustive search of the keyspace.

Yeah, thanks!

Nah, the Reddit-alike thread-based intercompatable forum systems on the Fediverse. Lemmy. Kbin. Mbin. Piefed. Sublinks.

Is this that girl who has the big archive of videos of robots that she’s made malfunctioning in funny ways?

EDIT: No, I don’t think so. This one has a short archive of projects on YouTube:

https://youtube.com/@shebuildsrobots

The one I’m thinking of has a large collection of them failing in funny ways.

I kind of wish that they’d unify some of their features.

For example, I use the “Fediverse Forums” search lens to search the Threadiverse only. That’s a drop-down menu.

Then there are those Duck Duck Go-style alias things that start with an exclamation mark.

kagis

Bangs.

And now the snaps.

Like, is it necessary to have all these as separate, segregated features? They all kind of do the same thing, are a way to ask the search engine to interpret the query differently.

EDIT: Also, I don’t know if there’s a Kagi lemmy community, but if so, that might be a better place than [email protected], since most folks won’t be using Kagi. Doesn’t matter much for communities that are desperate for traffic – like, for games, I’d rather talk on a general games forum until traffic hits some point, rather than having a lot of game-specific communities that are ghost towns. But [email protected] is one of the largest Lemmy communities, probably has enough post throughput.

I’m not a mod, not saying that it’s community policy, just thinking about where it might best make sense.

EDIT2: Looking at lemmyverse.net, there is, but it’s on lemmy.ml, and I’d really rather not subscribe to .ml communities. Doesn’t appear to be any other Kagi communities at the moment.

Well, I don’t really want to mod one myself, but if anyone wants to run a Kagi community somewhere off .ml, I’ll subscribe.

Apple’s study proves that LLM-based AI models are flawed because they cannot reason

This really isn’t a good title, I think. It was understood that LLM-based models don’t reason, not on their own.

A better one would be that researchers at Apple proposed a metric that better accounts for reasoning capability, a better sort of “score” for an AI’s capability.

There are probably some games that this would work well for, but I’m not sure that it’d be a great replacement the way a physical thumb keyboard is for texting or the like.

Most present-day games that I can think of that I play use the keyboard as a grid of buttons. They expect to have your hand over the thing – often the left hand, with the right on the mouse – to let you be able to push multiple buttons quickly.

I’m not usually doing much text entry, which is what I’d expect a thumb keyboard to work well with.

That’s a thought, though my guess is that access time constraints for something like a CD might mean that it could at most be a secondary form of storage.

The Wayback Machine’s site has been breached, but its founder says the data is still there.

One concern I do have that’s maybe worth considering is that The Wayback Machine is often used as an authoritative source of what a website was like at some point. Like, if you’re citing information, it’s considered appropriate to link to The Wayback Machine.

There are entities who would potentially be interested in being able to modify that authoritative history.

I don’t think that that’s likely an issue here – someone who wanted to secretly modify the history probably wouldn’t have also modified the site to indicate that it was compromised – but the ability to modify such a representation might have a lot of potential room for abuse.

It might be worthwhile, if the infrastructure permits for it, to do some sort of storage mechanism that makes it hard to spoof old data.

If you’re familiar with blockchains, they leverage a chain of hashes, so that there’s a piece of data dependent on all prior entries. That sort of dependency didn’t originate with blockchain – the CBC cipher mode does the same thing, off the top of my head – and I don’t think that a fully-distributed mode of operation is required here.

However, it might be interesting to use some sort of verifiable storage format where hashes of checkpoints are distributed elsewhere, so that if someone does manage to get into The Internet Archive, they can’t go fiddle with past things without it becoming visible.

Git repositories take advantage of this with their signed commits and hash trees.

If someone gets into The Internet Archive, they could potentially compromise a copy before it gets hashed (though if they supported the submitter signing commits, a la git, that’d avoid that for information that originated from somewhere other than The Internet Archive). This can’t protect against that. But it can protect the integrity of information archived prior to the compromise, which could be rather important.