How many people will die in car accidents trying to fix their audio....

Exponential growth evangelism pushes stock prices up while consumers quality of life goes down. Bad economy, happy shareholders.

We have reached the point where fighting this at the individual level is nearly impossible

result: quality goes down, previous partners get pushed out of business, new players struggle to enter the market, innovation slows

There’s always visible degradation in services/products when a provider decides to become “everything.”

Retailers pushing their own branded products, apps closing integrations and rebuilding third-party features themselves (Claude, X). Streaming platforms producing their own content (Netflix)

> to read long form blog posts anywhere in the app

Which means hard-coding content lexicons for each standard.site publishing platform?
The actual content is free form in the spec, and while pkct/leaflet/offprint have many overlaps in lexicons - others might not

And LLMs won’t fix this - they’ll make it worse.

“Our LLM rewrote a C compiler”
“Our LLM built a new browser”

Even if that software eventually becomes usable, it’s still just another pile added to the mess we already have

Thousands of frameworks. Thousands of incompatible APIs. Auth systems full of foot-guns. CLIs nobody remembers how to use.

We can’t seem to build tools that are as simple as going to an LLM and saying: "I need you to do this"

Hot take on LLMs: they exist largely because humanity has failed at systems engineering.

We keep producing half-baked standards with too much room for interpretation. Then we ship new fixed versions with backward compatibility. Then another standard appears

And the cycle repeats over and over

feature request for Chrome DevTools @developer.chrome.com

"Readers are burning out"
Great talk from @aendra.com on #AtmosphereConf, but algorithms and feeds are't the answer to everything
You shouldn't rely on algorithm to deliver news on topics that are important for you.
The whole idea of news delivery must be reshaped into targeted, focused streams

found bug - jump can't be canceled mid-air by pressing down unlike in the original

Post-credit scene? Nah, we’re waiting for the post-credit AI training restriction notice

We’re building Logue because there’s still no humane, standard way to follow important events and topics without getting dragged into attention-farming machines.

Lifeworld, by Oneohtrix Point Never from the album Tranquilizer

oneohtrixpointnever.bandcamp.com/track/lifewo...

That’s why distributed, many people sending one request will not qualify as unfounded or excessive

DDoB (Distributed Denial of Business) - a malicious attempt to disrupt a business by overwhelming it with a flood of GDPR data subject access requests, draining legal and operational resources

The eternal loneliness of a man wandering in a finished game

Ideal humane algorithm behavior:
- track post views only so you can notify people who viewed a post later proven fake (age-old issue: the headline hits the front page, the correction ends up on the last)
- base suggestions only on likes - so I control what I would see, not system guessing

big tech social apps algorithms don’t even need you to open a post anymore. Leave it on your screen a few seconds longer than usual and you’re pushed into suggestion hell.

And I really need "It was a mistake, forget I was here" button on youtube, tired of opening videos in incognito mode

#atproto could potentially become a future replacement for RSS. With projects like @standard.site and native permissioned data it could fix main RSS limitations like discovery, interactions, monetization

18EDO Bailiff, by Clark from the album Steep Stims

Microtonal music feels like a stronger drug
clark.bandcamp.com/track/18edo-...

Customizable interactive side panel for logues.. distant future plans

I never met a real fan of touch-sensitive buttons (maybe one new Audi owner, but he is just protective - these cars became a big touch-slop). And I don't remember when last time big corps did something lately to not just cut costs. If I can't find these people, where Apple finds them?

touch/rub/tap-sensitive interfaces (even with haptic feedback) being more cost-efficient to produce than dumb physical buttons - where progress failed us

And of course the whole project will follow @standard.site lexicons. Interoperability, openness, and decentralization are core priorities

Why do we have all these LLM benchmarks if all people want is just a pelican riding a bicycle...

So the best path for great ideas is: implement it, get users, evolve, break, rebuild, break again and repeat until it is perfect and then hand it over to W3C...

And then there’s ActivityPub. When you’re designing a social networking protocol it probably makes sense to involve companies that actually run social networks at scale - but that wasn’t the case.

OpenID Connect as example. RFCs had so much room for open interpretations that every new provider you add ends up coming with its own quirks