Stop thinking about how you are going to play games without RAM and start thinking "How could Sam Altman benefit from the 2.7 liters of blood plasma I am selfishly hoarding in my body?"

If we can't make them use Rust they can't make me use Linux.

I just want to say insanely funny bit from Broadcom's CEO. The man is a great stand-up comedian.

Thank you for opening the earnings call announcing a 143% YoY increase on AI-related revenue, just to go "Oops, sorry! Well, let me restart, guys..." after noticing that was from FY25Q2 and the revenue from FY26Q2 immediately sent the entire tech industry to wipe a trillion and change off the stock market.

Big womp womp for Amazon's satellite deployment failure today setting back some shit corpa's constellation. God willing Ariane 64 next week will explode mid flight and spare us from Amazon's own constellation too. I pray for nothing but the worst to befall the entire private space industry.

GitHub's Copilot generates the PR from Anthropic's Claude generated work plan, a Cursor BugBot agent generates a risk assessment report for the PR and a Vercel AI Agent reviews the PR, which is automatically merged immediately and without further actions by a human.

The CI that runs on the main branch on push fails to build. Inspecting the logs reveals that this happens because Vercel approved a PR containing an unexistant method reference which Copilot copied as-is from the work plan generated by Claude, which does not describe a functional solution to the prompt by the human engineer.

Further investigation reveals that Cursor was correctly omitted from the root cause analisys due to being wholly useless. The generated risk assessment was deemed ignored by either Claude or Vercel's context windows. It was further deemed completely unreliable, providing a list of claims both exagerating security checks and undermining possible exploit paths, the majority of which did not exist at all anywhere across the PR or the existing codebase. The generated summary classifying the PR as "low-risk" was additionally false if compared to the generated assessement.

After fixing the bug manually and restoring CI functionality, QA identifies the new feature as completely non-functional.

A second manual review is carried out, identifying several methods which Copilot failed to generate. These would be responsible to implement core functionality. A further dive into a true root cause leads to a split in blame between the code generated by Copilot and the work plan generated by Claude. The former generated non-functional code, while the latter generated a work plan without critical work items specifically declared by the human engineer's original prompt. The work plan instead was generated replacing such requirements for more efficient, less demanding steps, which lacked any technical detail and in some case where simply handwaving the problem away.

The code is deemed unusable, the PR is posthumously reverted from the main branch and a new PR is opened. The new PR contains a much longer and more detailed work plan generated by Claude based on a prompt written by the original human engineer, at least three times the size of the first PR's prompt. After 4 hours, the code is generated and ready, Cursor generated a risk assessment and deemed it "low-risk", Vercel automatically approved the PR, which is merged immediately to the main branch and CI fails to build it.

What do you mean "bottom of the barrel" buddy? Which bottom? Of the barrel on top or the one below it?

✅ The Regime knows best! 🫡

We need more bouba games and less kiki games. We need more nephew games and less unc games. We need more bro games and less fam games.

Elden Ring is kiki, but it's unc. Hades is kiki, nephew and bro. Isaac is nephew in Brazil right now, but it's unc everywhere else and it's also too kiki to be bro. The Plucky Squire is nephew and bro, but it's very bouba.

No further comment will be provided. I said what I said and I'm objectively right.

We have seen an explosion of successful and fun uses of RNG in 2025, following an already warm reception to several RNG-centric games in 2024. Is that a good or a bad thing for 2026? Is there still hunger for exploring RNG or have we sort of hit the limits of what is reasonable?

I'm mostly looking at how we started to see players get really hyped in 2024 about games like Vampire Survivors and Balatro, which basically started the "gamble-like" or "scoreslop" or "points-go-up roguelike" or whatever you want to call the genre. Then in 2025 we have have seen a flooding of the market with RNG increasingly at the center, alongside mechanics to rig it.

What is also happening, towards the end of 2025, is that we started seeing reasonable success for comical takes on RNG like Rogue Light Deck Builder, Unfair Flips and Q-Up. Usually those are early signs of a genre being pretty saturated.

With Megabonk's release and after months of RNG-centric games, players seem to be starting to hint at being burnt out, not a lot, but enough to see people complain that they are tired about what are objectively well designed games. There seems to be an increasing feeling that well designed is not going to be an automatic win anymore and the more games come out and the longer people play them and build up some burn out, the harder it gets to win them over.

I'm wondering, for 2026, what should we expect or want to play? What should be the focus of development efforts?

Is the way for Indies to keep exploring wilder ways to use RNG, or maybe step back a bit, reset to a tamer, more tried and tested design like the ones from 2024? And at that point do you go even further back, do you go to Isaac instead of Balatro, maybe Isaac or Spelunky and move back from RNG as an interactable mechanic that players can directly control?

Is there a place to go with no RNG at all? Find new mechanics to exploit, Subnautica had success with it, Ultrakill is still going strong. Are people maybe hungrier for fully controlled designs than they would be for RNG?

Maybe we haven't seen the whole of RNG either, the subject is so broad hat we most likely did not. Maybe we're tired of RNG roguelikes, so we could look at RNG RPGs where Dwarf Fortress is still insanely popular, so is Baldur's Gate 3 for example. Are people hungry for other kinds of uses of RNG, where RNG is more permanent and closer to use in worldbuilding
and storytelling than it would be to gameplay?

These are open questions. I don't have a firm answer on what I want and most players don't seem to have a clear idea either. It's something that has been bothering me for a few weeks as someone who exists in both sides of the gaming industry but regardless of which hat I wear I can't answer any of those questions in a definitive way.

It's an interesting dilemma for others I talked to, both developers and players, and while nobody has a clear answer either, the presence of a dilemma has interesting implications on the existence for a desire for change even if nobody knows what that may be, leaving this post as an open-ended and largely unanswered question.

happy friday afternoon

Asking Santa for a shitload of platinum I can sort of understand, but asking for THE MINE is fucked up! Also what the fuck is Santa doing holding South African platinum mines just ready to be gifted away to whoever asks? THAT'S NAUGHTY LIST BEHAVIOR SANTA STOP THAT

For free

In Finland we homo

"You call that a bubble? That's not a bubble. This is a bubble" *Crocodile Dundee pulls out an investment portfolio full of Nvidia, Microsoft and Oracle stocks*

scallywags found dillydallying in the steel brig will be made to walk the plank

if you have to use klarna for this you fucked up, hard reset time, go to the 6th or 7th floor right now

klarnaing the fucking 16€ driving license renewal fee into a 12 months loan at 70% interest

steam says 114h played in the last 2 weeks, that averages out a little over 8h per day, this cant be healthy

⚠️ Stop using anti Fedora Flatpak language!

Instead of saying:
❌ Why does Fedora Flatpak exist as a default when it provides no practical benefit to a user and has actively proven to break their own packages? 🤓

Say:
✅ The Regime knows best. 🫡

I chose not to choose Ads. And the reason? There are no reasons. Who needs reasons when you've got fucking uBlock.