Someone made a pretty good, albeit very cautiously, almost concern-trollisly written blog post about recent internet drama Odin v. Wikipedia. Surprise: the conclusion is that Odin is fashTech. (warning, the blog appears to be LLM-designed. I do not believe the actual text is LLM-written but you can never be sure with that kind of writing style)
https://katamari64.se/posts/2026/odin-wikipedia/
It can only do the impressive numbers after you give it a schema for the data, and the data has to be structured in a way that follows a rigid schema, so it’s not magic.
OpenZL compresses data into a self-describing wire format, any configuration of which can be decompressed by a universal decoder. OpenZL’s design enables rapid development of tailored compressors with minimal code; its universal decoder eliminates deployment lag; and its investment in a well-vetted standard component library minimizes security risks.
The freedom afforded by OpenZL’s graph model is a double-edged sword: the capability it offers comes at the cost of a combinatorial explosion of choices that must be made—namely which components to use and how to compose and configure them. There is no one-size-fits-all compressor
Webassembly for compression or something, but it’s made by the zstd people so maybe it will turn out to be magic eventually.
On one hand the x2.06 ratio for the SAO data is extremely good, zpaq and paq8px are 50x and 10000x slower at encoding and decoding and still can’t even reach that ratio (I didn’t try even more extreme compression, paq8px -8 already took 20 minutes).
On the other hand,
enwik: A Case Study Where OpenZL Performs Poorly
We present this case study to show that OpenZL is not a magic bullet for all use cases. The shortcomings are twofold. First, the standard codec library can achieve good results for many datasets, but is not currently optimized for human text. Second, English text is a highly compressible format for which domain- specific transformations are especially useful. xwrt combines multiple such transformations to achieve its results. With this in mind, we are optimistic that porting text-specific codecs and text parsing will improve the performance of OpenZL compressors in this domain.
The manager of the store says:
“Stop. You know nothing. You have baked 0 bread loaves by hand. No one has ever depended on your bread. You are a finger-wagging “they put dog poo in my bread” type in an era where you hide in plain sight coasting on the moral high ground of baking homemade bread from scratch at home without poo. You can’t ship bread, can’t adapt, can’t even realize that a bread shop is not the place for this kind of attitude.”
The free tier of IGDB has a rate limit of 4 requests per second and they will block you if you try to download the entire database, so this may be a dead end anyway.
Found a 4 megabyte dump that seems promising though it’s from 2018 so outdated.
https://gist.github.com/LeWawan/5858a9e7bef0f3dc4a79ac8bc1e3380c
Another one, a 68MB 3 year old “dataset” for AI training that seems related to IGDB.
https://www.kaggle.com/datasets/anudeepvanjavakam/igdb-api-data/data
The simplest fix is a delay between an update being pushed and the update being deployed everywhere. Several orgs are scanning all popular dependencies for supply chain attacks and they usually catch them quickly, just not quickly enough when there is no delay.
Doesn’t end there
The AI job apocalypse isn’t data-driven — it’s narrative-driven, engineered by people who profit when you’re scared. Fear is the product. Capital is the outcome.
When a resource becomes dramatically cheaper to use, we don’t use less of it — we find a million new uses for it. If that sounds painless, keep reading.
and that quote gets even more LLM when expanded
The AI job apocalypse isn’t an economic forecast — it’s a marketing strategy. We’re not witnessing the end of work. We’re watching the monetization of fear.
It’s two not X it’s Ys in a row!
Galloway has always been a prolific em-dash user But pre-2023 phrases that strike on the same severity are hard to find ——mildly LLM at best.
TikTok has 1.6 billion monthly active users — more than Twitter, Snapchat, and LinkedIn combined.
The new occupant’s ascent to the Iron Throne was financed with a different currency — not monthly subscriptions or cable packages, but attention. Specifically, our youth’s attention.
Competition depends on rules, and rules depend on umpires. We should fight to protect competition — not winners. Because winners subvert the process
But is it better than ncurses?
The incredible thing is this is actually the result of an explicit design decision.
The compiler accepts most GCC flags. Unrecognized flags (e.g., architecture- specific -m flags, unknown -f flags) are silently ignored so ccc can serve as a drop-in GCC replacement in build systems.
They’re so committed to vibing that they’d prefer if the compiler just does random shit to make it easier to shove it haphazardly into a build pipeline.
Cryptography is hard and programmers are notoriously really really really bad at it.
maybe they were looking for extra special characters like 🁄 or ⶸ. Who am I kidding, RFC 1738 tells us that literally everything is unsafe and you know, we need to prepare for the inevitable occasion when the password somehow ends up inside an URL.
The characters “<” and “>” are unsafe because they are used as the delimiters around URLs in free text;
the quote mark (“”") is used to delimit URLs in some systems.
The character “#” is unsafe
The character “%” is unsafe
It ends up with
Thus, only alphanumerics, the special characters
$ - _ . + ! * ’ ( ) ,
are safe
Rendering is not what you are doing server side.
No shit sherlock. Rendering requires information about the game, and that information is enough to allow cheating. Aimbots don’t need to perform “invalid actions” in order to wreck a game. They just need to be faster and more accurate than most human players. Trying to heuristically detect aimbots is also commonly used alongside other anticheat methods, it just doesn’t work (unless you have people manually reviewing individual reported cheaters, but companies try to avoid that because it’s expensive and risks false positives).
Ultimately you either have basically google stadia (with all its technical problems) or you are trusting the client to render the game.
Even if the client only has exactly the absolute minimum amount of information needed to draw all the things that are visible, that still allows a cheat to see the player coordinates and the coordinates of visible entities, which usually makes eg. an aimbot trivial to make.
“Never trust the client” renders entire genres of games inaccessible for a big corporation. But those genres have billions of dollars of potential profit in them. So they will go as far as they can to make the client almost trustable. The average player of a first-person shooter doesn’t really think about the implications of kernel-level anticheat at all so it’s not a hard choice for them. 95% of them are on windows after all and that already gives kenel access to their PC to some entity they really have no good reason to trust.
it’s
while (true) {
let t = Date.now();
if (timeoutMap.has(t)) timeoutMap[t]();
}
of course. Clearly O(n).
Feel free to use it. I guarantee it is bug free. Comes with express warranty. This notice is legally binding.
SteamOS for Desktop is imminent
Do you have some source? I recall the latest rumor related to a valve patent is more likely to be about some VR thing than a steamOS PC
“You want to use teams a bit? We have a session here” “I’d be happy to, actually. Not really, but it wouldn’t be bad” “Not really? If you say so, I have a teams session ready right here” “No. No. I’m not stupid” “People use it every day.” “Tell the truth” “It’s a good user experience.” “So are you ready to use it? For 5 minutes?” “No, I’m not an idiot.”
Also the point is to get attention of broader public, not just those at the labs.
The highest possible attainment, to generate several popular memes about crazy cult member does something slightly odd to show his devotion, but isn’t brave enough to do it outside his own home
memes often contain mild inaccuracies
They already have almost all of the discrete gpu market, they’d have to expand to new markets (although they are kind of exploring that already)
and with a good enough leak, the amount of unused memory will become negative!
Reading between the lines they wrote this because of seeing “I hear you, and you are right to be suspicious — but I only use AI for spell checking, nothing more” one time too many.