kasey_junk

You can get that working for yourself trivially. My agent environment now has 3 different kinds of compute abstractions, 3 different agents and a nascent workflow engine in it.

It works how _i_ want, which is really refreshing.

Could be. But a) I don’t know why I’d trust them to do that when existing tooling means I don’t have to and b) this is not a new paradigm for me. I’ve been running most of my dev environments in sandboxed mode for a long time. I don’t trust npm either.

The first step is building a sandbox (vm, container, whatever) that allows you to trust the agent in its most permissive mode.

This varies based on the maturity of my projects but most of my sessions live in VMs with very specific egress rules and permissions.

Pretty straight ahead coding workflows. Most of my operational workflows are much simpler (because diagnosing an operational issue is much more time sensitive obviously).

Yeah absolutely. And from other context windows for agents.

I have workflows that run for 8-10 hours unattended regularly. This is on frontier models though.

But if I were to look in on them they’d be giving feedback during that time, but they wouldn’t need steering.

I don’t know what the op meant by manually picking but the expert systems of the 80s and case based reasoning systems of the 90s used fairly static decision functions and were explicitly called AI at the time.

You can see shows there, which is the right way to do it.

I think the environment thing is really overblown as well.

What’s not overblown is the ease with which the government can get big projects done. And there are some downsides to that, specifically when it comes to property rights.

The Beijing public transit trains are a great example. The routes are wildly straight. So straight they allow extremely long trains.

Thats possible, in part, because the government can just reappropriate the land in a way that governments in the west can’t without engaging with a contentious public.

And that’s just one example. That sort of thing permeates many different kinds of projects and leads to a particular kind of efficiency.

A more charitable reading of this comment is that you can make a game theoretical argument for these behaviors that doesn’t require altruism. And that argument therefore doesn’t preclude going closed at a later date.

PayPal announced 5k layoffs in May.

The reason you know the people complaining the most about this aren’t serious is that they don’t lead with crsp and vti.

They did change their rules, they did it fairly specifically for spacex and it did drive inclusion in a major index fund (perhaps the biggest one).

Now me personally, as a holder of vti I am good with the change and my included exposure to spacex. Further I think mostly complaining about the inclusion/exclusion of a single name in an index _defeats the point_.

But for those decrying the shenanigans crsp and vti are the example to go with.

It’s not, and calling it so is wrong.

But it is a country with extremely illiberal capital controls and a history of questionable enforcement due to perceived corruption.

It’s such a good payment system it’s overcome these concerns in large part, but they are reasonable concerns to have.

Yes? That’s what the whole movement is about in the US and it’s the lead reason dua lipa quotes in the linked article.

Isn’t the American military logistics the most decentralized supply chain in the world? Famously (perhaps apocryphal) _every congressional district_ has jobs in the military logistics supply chain.

They are banned though. Not curated. You are not allowed to have these in school libraries. It’s not an editorial decision. For instance the Utah law says that a book must be removed from all libraries if 3 school districts in the state ban it.

No librarian, or teacher, or admin or parent in the other school districts gets any say.

That’s a ban. People may not like that their state is engaging in authoritarian behavior, and it’s less authoritarian than other behaviors, but it’s a ban by the simple facts. Doublespeak doesn’t change that.

Yet most people in the watch industry will suggest that the Apple Watch was a boon for the industry because it retrained people to wear a watch, a fashion that was being abandoned.

If you don’t allow something to exist in a school or classroom library through statute or governmental action what do you call it?

Would you prefer forbidden? Barred? Censored?

The banned books movement (on either side) is broadly not about curriculum. It’s about access in non-required spaces such as libraries and clubs.

The issue there isn’t that kids are being exposed to these items, the issue is that other parents are censoring what _my_ kid can be exposed to. They are infringing on _my_ rights.

Meanwhile I’m not requiring their kid to go into a library and checkout Maya Angelou.

Your access records are only as good as your ability to prevent someone from making unauthorized access.

It’s also an usual use of the term arbitrage

It’s probably important to define what sort of code review you are talking about when making broad claims about it.

GitHub style asynchronous pull request review with inline comments is the norm now, but it’s not the only sort of review there is. I’m old enough to remember processes that include in person reviews that were more like a dissertation defense or conference presentation.

The literature around this that shows that code review is a useful quality practice (in fact one of the only useful quality practices) comes mostly from much more structured review processes than we see now.

My personal opinion is that before llms the GitHub style pr review was for making us feel better about our processes (or governance checkbox checking) and the age of llms will sweep them away as the cost/benefit is so much worse now.

Their harness is terrible compared to any of the other cli based harnesses I test against. Like shockingly bad.

This comes up all the time at work because the vendor management people don’t understand the llm ecosystem and think Claude through copilot is the same as Claude through Claude code.

A simple side by side comparison will show dramatic under performance 3 or 4 times out of five when I’m asked to explain the difference.

One obvious distinction is that Tarrio was convicted of leading & planning the Proud Boys operation on Jan 6 while this defendant was convicted of moving zines. Tarrio wasn’t at the insurrection on Jan 6 because he’d already been barred from traveling to the city…

And he got less of a sentence. I don’t think your argument equating these 2 is arguing what you think it is.

If your hiring pipeline is employing a filter that a) is not better than a random chance and b) is expensive to implement get rid of the filter.

Instead of spending all those resources on resume filtering, hire resume blind. Instead of using llms for a thing they are bad at (subjective decision making) use them to build a deterministic process that isn’t.

Use work sample hiring as the filter. Make the work sample automatic to sign up for and judge.

In the simplest incarnation I’m just using the built into the agent cli parm’s that trigger non-tui behavior (for instance calling codex exec instead of just codex).

Most of my workflows have slowly moved away from the chat interface with llms. Instead they look more like traditional Unix pipelines that just happen to call Unix tools that interact with llms.

This allows me to make more repeatable processes, not be tied down to vendor implementations of workflows and mix and match models for cost and efficacy.

There is nothing that ties you to talking with the text generator black box, and for most of my use cases it’s a negative.

That’s already an option? You can get small business loans, bootstrap, get grants etc that don’t require massive growth.

You can found or work for a company like this anytime you like. But the “leeches” the op mentions are a voluntary funding mechanism for a particular kind of company. If you found or work for one of those the trade off is clear.

You can’t have it both ways though. As an employee you can’t live off the largesse of investors as you build the business and then not expect them to want an elevated return on that risk.

The reason hockey stick growth is required is because the “leeches” are putting up the capital to build the profitable company and if the profits dont significantly outpace the risk free rate it’s a very bad investment.

The founders and employees and even the customers are accruing all the benefits of that capital so of course they are happy.

How else do you propose funding the quite expensive and risky enterprises that venture backs? Taxes? Paying employees less before profitability? Charging early customers a lot more? Clearly you can see the downsides of those approaches.

The preorder delivery window they offered me was effectively q2 2027. No idea what the likelihood of them delivering it by then is.