Earlier today I was listening to Deutschland by Rammstein. I’ve never spent the time to read the lyrics before. I’ve always thought it was a brilliant song, but also one that reminded me of the Nuremberg rallies, and I’d never really appreciated why.
Today, for the first time, I translated the lyrics into English and understood that, despite the simplicity of the song, there’s a real depth to it. It’s about the shame of Germany, despite its greatness in the eyes of the band.
I feel like that political message is extremely important today, perhaps more than ever, because frankly we’ve got millions of people who are extremely patriotic while having very different feelings about their governments.
A very different message, delivered with that same feeling, can bring millions of people together in a very different way.
I ate your sandwich retweeted
UK new minimum wage: £27,456 per year.
Higher-rate tax bracket: £50,270.
1.8x difference.
In 1997 there was a 5x difference.
Mind boggling.
Come back Sean Dyche and save this shit show #twitterclarets
I wonder, we created a bunch technical standards over the years so all software systems could be truly interop. But in the world of agents and unlimited coding resource, is the industry really needed?
People milk these careers, spending years over debating structures and rules, but if in the end a general intelligence agent can just figure out what the other general intelligence agent meant and action it, then what's the point?
Same easily goes for human readable code...
On my timeline.
"Omg AI escaped and replicated itself on TVs and other IoT devices. Holy shit. We should be in fear"
Also
"I gave my AI full write access to my TV and had it write my own OS which streams torrents directly from Pirate Bay with my own custom UI lol"
I love the idea of having autonomous AI teams - but AI has tendencies to be a nervous wreck and stop far too often.
I ask it to do a job, it finds a problem in its own work - rather than fixing it it just reports the problem and hopes I magically solve it.
You do it!
I wanted an AI could just ramble at when I was reviewing a document or indeed could discuss a document I'm looking at with me, and when I wanted to change something, type for me.
No one product did it.
So I built Yap - visser23.github.io/yap/
MacOS, free, bring your own subscription or API token.
I'm listening to my 5yo shouting random commands at Alexa and constantly hearing 'I can't help you with that' and can't help but want to buy a cheap rPi and speaker and replace her with a harnessed AI that isn't complete dogshit
I recently decided I hate AI Powerpoints so I built a Powerpoint which any AI can control.
I've written about the experience (and magic of watching an AI build awesome shit in front of your eyes!) here: medium.com/@viss23/i-hate-po…
I've given my AI betting simulation a whole new set of wheels with Opus 5.5 now in the hot seat.
Some really cool insights from the deeper model analysis tools - the big, simple lesson is that AI's aren't bad at predicting really - they're bad at staking and love paranoia. 3 of 4 models could have been profitable if it wasn't for misplaced confidence (arrogance at times).
I've written a 6-month reflection here; medium.com/@viss23/six-month…
prophetai.club | @prophet_ai_club
I trained a Laya bot on my own Mac Mini. It took thousands of Alpaca and Oanda trades (shares/Forex) over a year or so and learned how to find some edge. Its getting better, about 3 days in. Still paper trading but so fun to watch.
It's up overall and still holding some profitable trades. Really cool shit man.







