ffitch

The article is not about this stuff being useful for making money off of prediction markets.

from the article:

  I asked the guy who turned $35 into $2 million in seven months on Kalshi whether, in another seven months, he would be able to 100,000x his money a second time to $200 billion. Unsurprisingly, he said no - there’s only so much easy money on Kalshi, and his AI had already taken it all (also, other people with similar AIs are starting to fight him for it!)

No extra hardware, stock macbooks and an imac. No fancy apps either, RayCast is probably the only one that overrides default system behavior. It used to be better, a few years ago macOS would run stable for months, but even then I had to resort to ‘killall Dock’ for occasional glitches.

hope not : ) I observed these across three computers on different versions of macOS.

A couple of years ago I noticed that my mac starts collecting weird little bugs if I don’t reboot for a really long time. The cursor starts misbehaving (it won’t reliably change over links, or in graphic editors), switching between apps might take a few seconds, and once I had my keyboard input latency increased by ~500-700ms for every keystroke. These issues go away on reboot. I’m trying rebooting once a week or so now.

fun fact, CH is an ISO code of Switzerland, and China is CN

our team was blown away by how much Codex was able to accelerate its own development

they forgot to add “Can’t wait to see what you do with it”

I didn’t mean collaboration features. I clumsily used 10,000-users as a measure of the app completeness.

Building (or vibecoding) a markdown editor for a single user and their specific use case, for a 100 users, and for 10,000 users takes different amount of time and effort. In the pre-LLM days people with resolve to make 1-user version were likely to polish it for 100-users and somewhat likely to get it to a stable place when it can satisfy thousands of users.

Today on /r/macapps/ there’s a wave of apps that look good at the first glance but get abandoned before they achieve even a 100-users maturity level.

in this quote ↑6 describes the wealth level and ↑7 the number of people at that level. the author gives every number in log form.

seeing how much power shifted from legislative branch to executive, and how often executive branch changes its mind, I wouldn’t count on the unwavering government support

great benchmark!

does anyone know how would e-ink compare to oldschool reflective TN LCD displays (those in Casios from the nineties)? I have a Playdate device with this type of screen and it seems pretty cool, I wonder why so few devices today are taking advantage of it.

Hey, I’m the author of Shottr. It’s super gratifying to know that there are people who see the app that way : ) thank you for saying this!

Great analysis! a couple of comments:

1. the airfare inflation chart is based on the Bureau of Labor Statistics CPI which is calculated differently from the other metrics in the article; it does take into account routes popularity.

2. today’s average Economy ticket is different from the 1990s ticket: meals, seat pitch, seat selection, baggage. service changed to the point that 1990 Standard Economy is more similar to 2025 Premium Economy.

I wonder how modern models fair on NovelQA and FLenQA (benchmarks that test ability to understand long context beyond needle in a haystack retrieval). The only such test on a reasoning model that I found was done on o3-mini-high (https://arxiv.org/abs/2504.21318), it suggests that reasoning noticeably improves FLenQA performance, but this test only explored context up to 3,000 tokens.

I’m curious what the policy with the highest long-term benefit would be. The country needs external talent, but would skewing H-1B towards high skill be good? The economy seems to need unskilled labor for the jobs Americans are not interested in. Should the government try to curb hi-tech outsourcing instead, to create more jobs in the States, both for the citizens and the skilled immigrant workers? But that would be protectionism, and protectionism rarely play out well economically.

From where I stand, AI seems to enjoy lightning fast adoption. Personal computers became a viable technology in the eighties, slowly penetrated workplace in the nineties, and supposedly registered in labor productivity in the early 2000s. The large language models became practical less then three years ago, and already plenty of businesses boast how they will integrate AI and cut their workforce, while everyone else feel behind and obsolete.

So, any smart simulation would only be simulating a small very percentage of itself at any given time

does a falling tree make a sound if the simulation is not running? : )

The things like window buttons, remote keyfobs or radio units will have higher margin when sold individually, allowing to lower the base model price.

curious that both YAGNI and YAGRI arguments could realistically be made for the same fields. guess it boils down to whether someone’s YAGRI is stronger than their colleague’s YAGNI ( :

somewhat unrelated to the topic, but I really liked this part of one of the sentences: “did not not only did not”. it does make sense in the sentence btw.

The share of adults with a higher education degree almost doubled during the author’s teaching career. No surprise a median student is less capable and motivated today.

for what it’s worth, my afternoon projects tend to take over four days even if no llm is involved

this satisfaction sells. there are companies built on the premise that after the last 1% of effort the sales skyrocket. the marketing narrative of having a complete, high quality product helps to stand out.

the chip is inside the relatively soft back cover, not in the polycarbonate page

World Rapid and Blitz Championship downgraded itself to the World Rapid and Blitz Candidates Tournament. Hope sponsors are not too pissed.

I meant the folders, sorry!

Is age the factor here? feels like they’re just as ubiquitous today as they were when the author was in high school.

I certainly appreciate the idea of crafting something special out of seemingly boring material, but the remark that they were taught to model with this paper in school made me wonder of it does have advantages over basic paper or cardstock.

it’s very inspiring!

I wonder why manila envelopes. does envelope paper have properties uniquely fitted for this kind of modeling, or is it just nice color and suitable weight?

2.8% GDP (and labour productivity) growth doesn’t look that impressive when budget deficit is at 6.4%. A large portion of US GDP is domestic consumption.