zeitraum.blog

tl;drthe new art is designing systems. time is the oldest example: four thousand years of the babylonian hour, and nobody asks questions. on mars you'd have to start from scratch, and suddenly beats become the safer engineering choice over stretched mars-

It is easy to imagine a hundred ways in which AI could go wrong. But there is one concrete area where AI can genuinely benefit humanity: rescuing science from its current crisis.

tl;drscience is in a systemic crisis: papers as kpis, broken peer review, publisher lobbying, attention economy. the result is a mass of things that look like science but don't withstand scrutiny. fewer than half of psychological studies can be replicated

all the building blocks for the next ui revolution are on the table.but nobody has put them together correctly. that reminds me of something.

tl;drthe history of human-machine interaction has seen two major breaks: the mouse (1968-1995) and the smartphone (1997-2007). both followed the same pattern: the technology existed long before the right interface was found. we are at the beginning of the

chrome 146 contains an api that almost nobody has noticed yet.it will be the beginning of a fundamental change of the internet.

tl;drWebMCP is a new browser api (chrome 146) that lets websites expose structured tools for ai agents. instead of screen-scraping, agents call functions directly. this shifts interaction from surfaces to interfaces, marking the beginning of a fundamental

ai-assisted development tips the build-vs-buy equation: why custom software is becoming economically viable for mid-sized companies for the first time, saas is losing its economic moat, and the new core competency isn

tl;drai-assisted development makes custom software economically viable for mid-sized companies for the first time: faster development, dramatically cheaper maintenance, software that adapts to processes instead of the other way around. saas is losing its