I read tons of comments like "It's not [this], it's [that] instead!" which is also wrong.
The performance bottleneck is whatever resource hits saturation first under the workload you actually run: CPU, memory bandwidth, cache/allocations, disk I/O, network, locks/coordination, or downstream latency.
Measure it, prove it with a profile/trace, change one thing, measure again.
From the FAQ in the "about" section of the website:
- You can try to find Wally (Waldo), as in the well-known game Where's Wally. He is partially visible, but if you click on him, he will appear in full and wave at you.
- Quest #1 - Mafia Treasures. To start the quest, go to the room where the Mafia is holding the annual meeting and click on the suitcase.
- Quest #2 - Subspace Tuner: To start this quest, click on the large advertising screen that says 'Bad Signal' next to the pirate ship.
- Payphone - you can call different subscribers on the 796th floor. Subscriber numbers are constantly being added and can be found in various places on the floor.
- In the Police Station click on the big screen to see the project statistics: current online, visits by country, number on interactions with all elements on the floor, etc.
- One of the arcade machines has a real game - Racer796.
- In the park zone there is Change My Mind guy. Click on him to add your own phrase to the rotation.
- You can compose a 10-second melody and add it to rotation by clicking on the guy in the hospital with the pink synthesizer.
- You can draw small pixel animation and add it to rotation by clicking on the Fun Drawing Screen near the Police Station.
- Click on the Chunk Norris in the park zone.
- Click on the JAWS 19 ad screen in the block with Back to the Future references.
- Click on Naruto near the pirate ship.
- There is Free Ads Board next to the pirate ship. You can draw your own advertising screen and specify which URL it links to.
- There are also many small interactive elements on the floor, clicking on which will show an additional picture, play a sound or cause some action to occur.
I'd prefer a 30 minute response from GPT-5 over a 10 minute Response from {Claude/Google} <whatever their SOTA model is> (yes, even gemini 3)
Reason is: while these models look promising in benchmarks and seem very capable at an affordable price, I *strongly* felt that OpenAI models perform better most of the times. I had to cleanup Gemini mess or Claude mess after vibe coding too much. OpenAI models are just much more reliable with large scale tasks, organizing, chomping tasks one by one etc. That takes its time but the results are 100% worth it.
Hacky workarounds aren't rare exceptions; they're the plumbing of modern software. Anti-cheat and antivirus tools only work because they lean on strange kernel behaviors. Cloud platforms ship fixes that rely on undefined-but-stable quirks. Hardware drivers poke at the system in ways no official API ever planned for.
Yeah, they're ugly, but in practice the choice isn't between clean and hacky; it's between shipping and not shipping. Real-world software runs on constraints, not ideals.
On the other hand, everything you ship outside of a clearly established golden path is a maintenance burden that piles and piles and piles. And these maintenance burdens tend to gradually slow the org down until they cause rather catastrophic failures, usually out of security or hardware (read: fire) incidents. Or HR reasons because people figure there are better places to fight fires.
The performance bottleneck is whatever resource hits saturation first under the workload you actually run: CPU, memory bandwidth, cache/allocations, disk I/O, network, locks/coordination, or downstream latency.
Measure it, prove it with a profile/trace, change one thing, measure again.
reply