bnchrch

Python needs to be in less places not more.

I believe Python has done more harm than good to the industry.

Its by nature very inefficient, it leads you in the direction to some very poor abstractions (inheritance, monkey patching, mixins etc) and as a result code that is very prone to subtle bugs and issues.

It deserves all the love it gets as an approachable language, but it deserves a lot more flack for everything else.

Happy to see Swift progress. Its a beautiful language and well thought out. Would love to see it expand outside of iOS

Yes please, take all the bleeding edge tech stories out of the aggregator I go to for bleeding edge tech stories.

I don't think the "don't really save you money" hot take holds water in every case.

Coding, maybe.

But for operationalized/repeatable tasks it definitely does.

For example I have a workflow that I was running in April that effectively would cost $30k in token spend for each full run.

However now, with GLM 5.3-flash, we've brought the cost down to $7k-9k with our evals showing we've had no loss in recall, precision etc..

Not the OP but chiming in as someone who has worked remote almost exclusively since 2015.

In my opinion that's the crux of this debate: social needs vs social remedies.

If you have high social needs but don't have a strong personal social network, then an office is a great remedy. It solves the problem for you by default.

Remote only seems to work for the inverse. Lower social needs, or a strong social network outside of work, or some mix of the two.

I'm fairly extroverted (as far as engineers go) and I'm grateful to have a social network I've cultivated personally. So remote has fit well for me. I can get my work done to a high standard without the obligation, then turn around and spend purposeful time with friends and family.

It's perfect for me, but not for everybody, and you can see that in the comments here.

What remote really did was expose how many people were relying on office culture to meet their social needs. That's not a bad thing. It just surfaced something that went under-noticed for a long time.

Automobiles are not interesting or useful because they're justing using trails the horses already built.

An 11% jump over opus 4.8 and a 22% jump over gpt 5.5 on Agentic Coding Benchmarks is certainly impressive.

Obviously still need to verify it for myself to see if it's truely a leap.

But am I the only one wondering, "What can I do today that I couldnt do yesterday?"

Previously I would think "Oh I wonder if I can finally get it to do X now?"

However now I feel like yesterdays models were more that capable to handle nearly any engineering task I paired with it on.

Maybe this is the final leap where I can comfortable set up an autonomous coding loop? Maybe.

Oh man a few things

1. A dashboard that tracks my personal metrics (github, strava, todo completion, flossing)

2. A eink display for that dashboard

3. A realtime node graph that shows a codebase (and/or its diffs) in a way that I can visualize what functions call which, and under what conditions

4. A agent that automatically fills out government forms and creates invoices for my friends brewery based on the delivery notes in their google calendar.

Hard for one person to keep up with a product that ships updates multiple times a day.

I'm sorry. But whats the point of this critique?

That a raw LLM hallucinates?

That we never see all the mistakes and dead ends a complex system using AI hits?

Does it even matter if its accuracy rate across all its experiments is < 100% if it can run trillions of experiments in the same time a human could run 1?

We don't see many of the failed attempts of Human Researchers. Why? Because it doesn't matter.

What amazing here is that it shows our society can make discoveries faster in the post LLM world. Thats incredible.

Your "critique" of how it happened. Not so much.

Our teams efforts have shifted toward reviewing plans as a team.

Before code hits a humans eyes it's went through a few independant review passes from our review agents.

If it's low complexity / blast radius it gets auto merged.

If it's high complexity / blast radius it gets flagged for human review.

And funny enough our team has agreed even our human review layer has the best results if we have an agent create supplementary descriptions of the code and potential issues that we read and apply human judgement.

Reason being is that it can be very pendantic. Both in a good and bad way. Either flagging things that are non-concerns, or catching things my lazy human eye wouldnt have caught. e.g. A docstring mis-describing the actual shape of an object.

I'm not sure it's fair to call the Apple Vision Pro a flop in the traditional sense.

While it may not have sold millions of units and been a household staple.

It certainly focused the entire org on manufacturing a suite of chips and hardware that are on a completely different level than their competitors. Apple's now has a clear advantage in all dimensions that matter: compute, power consumption, size, capabilities, etc.

Apple Vision helped created a moat that will be hard for anyone else to cross for at least a decade.