I’ve read a few different experiences with hosts having had success with Anubis to cut down on excessive scraping. One that comes to mind is the Dolphin project.[1]

I’ve read a few of those threads, but often it’s people at cross-purposes with the goals of Anubis. Is there a chance you could clarify the “not working” bit?

[1] https://dolphin-emu.org/blog/2025/06/04/dolphin-progress-rep...

From a few weeks ago: https://news.ycombinator.com/item?id=49500040

Basically, the cost of an optimized solution is orders of magnitude lower than the cost of an in-browser solution. Anyone dedicated can easily afford to solve workloads higher than your users will tolerate.

You might stop casual scrapers, but you're not going to stop someone who cares. AI scraping companies care.

That's not even the fundamental problem. Even if the payload runs optimally in the browser, the cost of CPU is so small that it's basically irrelevant.

If you waste your user's time with something that would take a full minute to run on a datacenter core, you're costing the scraper something like $0.000005: 360 W TDP on a 128-core EPYC 9754 * $0.10/kWh. In reality, it'll be substantially less than that, because CPUs don't use 0W at idle.

The only way this would make any sense is if there were many more scrapers than users and scrapers cared more about latency than real users, but that's the exact opposite of reality. The entire endeavor is so fundamentally misguided that it almost seems like a psyop.

You're assuming that the part of Anubis that stops bots is the PoW. It's not.

The other parts are even more trivial for someone who cares to work around.

"someone who cares" is the part that creates the firewall.

With modern AI tools, the amount you have to care is slight. Not much of a firewall. And the bad actors causing these scraping problems care a lot.

Still have to care enough to run the ai

> Is there a chance you could clarify the “not working” bit?

I see anubis, 90% of the time I close the tab before it finishes.

But why?

Impatience and annoyance

I sometimes wonder if the verbosity gives the LLM more “margin for error” when generating text that passes requirements.

It's simpler. The verbosity, the line of code count, is what has gotten people excited about it. So everyone optimizes for verbosity.

This does not strike me as true at all. Nobody's crowing about the verbosity of their AI-generated code. Rather, it seems to me they're impressed by the functionality or capabilities relative to time/cost.

10 lines of checks per line of actual execution. Anxiety for their stable pay

Of course. Those 10 lines of extra code is that much more tokens every time your agents read and use that section of code. The more blot in the project to more tokens your agents will burn.

Reminds me of a subcontracting agency

Spector was in his mid-forties when Deus Ex was released, System Shock a few years before that.

I wonder if he’s an exception that proves the rule like you mention, or if his experiences in his youth working on tabletop games contributed to his design sensibilities in the 90s.

Part of me always felt the wacky setting helped with the player’s sense of anticipation.

Or, you’re always learning new things and encountering new elements, be it plot, interactions, or simply exploring. The replayability ends up feeling quite arcade-y too, trying for a particular outcome.

It’s also worth noting that the EA studio that makes FIFA has been continuously doing so for over twenty years. It’s not being handed off between studios and rebuild operational expertise over and over.

If anything, FIFA was probably getting an incredible deal from EA given the cost of making AAA games today.

Claire makes a good point about standards improving require people complaining about their faults. Those gripes affect the discourse!

I send around 20 a year. Checking the mail is relatively monotonous for people, so receiving one is like a lucky gacha pull in today’s world.

Since they’re so memorable, it’s usually an easy way to make someone’s day without much time, money or effort. There’s also very little scrutiny in what you write, so even an in-depth episode about sun glare at the bus stop goes a long way.

I think the big reason they get such good framerates is their conservatism with shader counts. A staff member mentions that in an interview with DigitalFoundry for Doom: The Dark Ages. I wonder if their in-house technology made such restraint easier.

Compared to other engines, making a new material asset is easy for someone less technical to do willy-nilly.

For sure, if you restrict what can (and can't) be done in shaders you can optimize buffer layouts and rendering passes, not allow things that explode shader permutations/switching or explode shader instruction count and kill occupancy.

The flip side is the artists loose control and cannot author custom materials.

> The flip side is the artists loose control and cannot author custom materials.

...which is a good thing tbh. No game needs hundreds or thousands of custom materials.

I think it can actually be a bit better in terms of artistic consistency to have a fixed set of materials you can work with in a given game. Very important if you want any sort of stylized rendering!

You’re certainly right, especially for the AAA games coming off the presses, but it’s not always true with certain indie games, especially if they’re doing a physical print run long after publishing.

You can occasionally get one where a very final version copies straight off a PS5 disc.

If that were the case, I think we would have already seen a turbocharged AI fork of Godot by now.

Or, if the machine-generated programming is that good, it shouldn’t be prohibitively difficult to surpass the handwritten version.