ed

They’re not.

> the highest paid eng right now is getting 600 base + single-digit percent equity

I own a model 3 but would be hesitant to buy another Tesla given there’s a genuine risk the company won’t be around in a few years.

Remember you probably were exercising until about 18, since it’s required by most schools.

Anyone old enough to remember EPIC 2014? It was a viral flash video, released in 2004, about the future of Google and news reporting. I imagine 2027 will age similarly well.

https://youtu.be/LZXwdRBxZ0U

Here’s a gift link to the source NYT article: https://www.nytimes.com/2025/06/13/technology/chatgpt-ai-cha...

Thanks! I’m a native speaker (US) and was unfamiliar with this idiom. Cambridge dictionary does list this definition of beard:

> to face, meet, or deal with an unpleasant or frightening person in a brave or determined way

Quicksilver[1], app switcher [2]… Sherlock[3].

Now that Apple knows what to build, they’ll build it.

1. https://en.m.wikipedia.org/wiki/Quicksilver_(software)

2. https://www.folklore.org/Switcher.html

3. https://en.m.wikipedia.org/wiki/Sherlock_(software)#Accusati...

It doesn’t. At least not in the guidelines or FAQ (links at the bottom of the site)

The patch format addresses all of these issues, no?

https://git-scm.com/docs/git-format-patch

Neat. Wonder how this compares to Segment Anything (SAM), which also does zero-shot segmentation and performs pretty well in my experience.

I loved lodge 49. It inspired me to attend a welcome dinner at my local freemason lodge.

Sadly I discovered first hand why membership is declining (this lodge was a magnet for socially inept conspiracy theorists).

Doubt. It’s a press release from https://www.flogen.org/?p=98#toop

> auto-inclusion and auto-exclusion of various tools via an orchestration agent

Where do you see that? That would be neat, I'm under the impression orchestration is manual though – you define an agent and give it the ability to hand off tasks to sub-agents.

I wouldn't really consider this an attack (Claude is just doing what it was asked to), but maybe GitHub should consider private draft PR's to put a human in the loop before publishing.

I spent a few minutes playing with Studio and the model and agent are very impressive.

But be sure to connect Studio to Google Drive, or else you will lose all your progress.

Interesting!

> a 2023 genetic analysis discerned such a human ancestor population bottleneck of a possible 100,000 to 1000 individuals "around 930,000 and 813,000 years ago [which] lasted for about 117,000 years and brought human ancestors close to extinction."

And relatedly...

> A 2005 study from Rutgers University theorized that the pre-1492 native populations of the Americas are the descendants of only 70 individuals who crossed the land bridge between Asia and North America.

> The Neolithic Y-chromosome bottleneck refers to a period around 5000 BC where the diversity in the male y-chromosome dropped precipitously, to a level equivalent to reproduction occurring with a ratio between men and women of 1:17.

https://en.wikipedia.org/wiki/Population_bottleneck#Humans

Yes. It costs me a few bucks per feature, which is an absolute no-brainer.

If you don't like what it suggests, undo the changes, tweak your prompt and start over. Don't chat with it to fix problems. It gets confused.

DeepSeek wins most benchmarks according to that chart, so not quite SOTA, and there's no mention of model size so it's hard to compare efficiency

Counterpoint: externally, at least, Shopify seems to be a well run organization and that has to come from somewhere. (Parter for about 8 years.)

Interesting direction for research but not a model you’d want to use today. The paper looks at a 3b model built on llama3.2-3b, modified for mamba, and they’re comparing to a distilled version of r1 with 1.5b params.

Yep, it’s clear that many wins are due to Llama 4’s lowered refusal rate which is an effective form of elo hacking.

It’s worth trying the demo - I uploaded a low quality video of an indoor space and got decent results

It’d be nice if this post included a high-level cookbook for training the 3rd approach. The hand-waving around RL sounds great, but how do you accurately simulate a customer for learning at scale?

Do you have a real world example of this? Claude Code for example doesn’t fit the pattern of “higher success but more variance.” If anything the variance is lower as the model (and tightly coupled agent) gets better.

Kind of crazy Kevin wasn’t required to mention it was his project!

For those who want to do something, what is the 2025 version of Indivisible?

One moat will eventually come in the form of personal knowledge about you - consider talking with a close friend of many years vs a stranger

Nah the extension API is pretty limited. Copilot uses proprietary API’s not available to extensions.

If you really want an integrated experience, and not just a sidebar UI, you need to go the same route as Cursor and fork Code-OSS (the MIT-licensed part of VS Code, analogous to Chromium for Chrome)

Folks say the same thing about the 1991 Oakland firestorm. If there's a fire in your area, pay attention and don't assume it'll be contained, the situation can change very quickly.

Ditto. This was my conclusion after spending a bit of time building https://robocoder.app/

Coupled with the fact that devs prefer open source tools and are capable of (and often prefer) making their own tooling, it never seemed like a great market to go after. I also encountered a lot of hostility trying to share yet another AI developer tool.

(Note I am one of those developers who prefer open source tools — which should’ve been a hint…)