I'm pretty sure all these models have terms of service that make the user assert that they have permission to use the content you're feeding into them (clickwrap infringement-is-the-user's-fault). This kind of integration makes a mockery of that.

Amazon is big and inconsistent enough that "somewhere in Amazon, <XYZ> is occurring" is statistically true, no matter how nutso-sounding your <XYZ>.

Can we have a list of AWS services which use formal methods and general sanity so I can avoid the other 500 half baked ones?

> If you work in construction, you need to lift and carry a series of heavy objects in order to be effective. But lifting heavy objects puts long-term wear on your back and joints, making you less effective over time. Construction workers don’t say that being a good construction worker means not lifting heavy objects. They say “too bad, that’s the job”.

My parents were both construction workers. There is an understanding that you cannot lift heavy objects forever. You stop lifting objects and move to being a foreman, a supervisor... and if you are uncomfortable learning to get others to do work that you before have done yourself, you burn out your body entirely and the consequences are horrible.

This is factual reality, but it is also a parable that has been important for me to internalize about delegation in my own career. It is not irrelevant to AI use, but I don't think it slots onto it totally as neatly.

Also worth noting that, so long as it's reasonable, lifting heavy objects makes you stronger, whereas the current hypothesis on using AI is approximately the opposite.

Software developers are more architechs than plain programmers. You wouldn't make an architech lift heavy things, you want they to design how those heavy things are used.

Young people have more poorly developed impulse control.

> To lambasting them when they express any desire to actually form a family.

What is this referring to?

This person took a different attitude toward it and it was fine. Do you know someone who was attacked for interrupting a set or something?

> We don't have height categories, we have categories based on sex.

I mean, we do have weight categories in combat sports, right? I don't see why we couldn't come up with similarly neutral categories if we think it's good to segment people out by physical advantages. The parent comment is making a good point, though: it feels like some people care a lot about physical advantages that map onto gender stuff they care about, and not a lot about weird genetic anomalies that provide physical advantages that aren't gendered.

We could do that! I'm just trying to say that given categories based on (biological) sex, we should find some criterion based on biological sex to sort people into said categories, which the OC decision seems to do (at least better than the alternatives I have encountered). I don't have a problem at all with finding different ways of defining categories for competitions.

Re: anomalies - I think this is just unavoidable in any sort of category system, and I don't have a good solution for it except to consider frequency and severity.

I am in the same boat. Reading is a transaction and lately everyone wants to put 60 seconds of effort into writing an article and expect me to put 10 minutes into reading it, and I just can't. The writing feels dead, soulless even. Every sentence or phrase is structured like a mongering, click baity headline and it's insufferable.

I'm always interested in write-ups when folks try new attacks on self-study.

I will also admit that this part hurt my heart to read (vicarious embarrassment):

> the recruiter mentioned I needed to pay more attention to code debuggability (whatever this means - I assume that under the corpo-language, they mean that I wrote invalid code)

I completely understand why that line caused vicarious embarrassment. Looking back, I realize my brain was(is) operating on a completely different definition of that word based on my daily constraints. I plan to write more about this in Part Two, but at that point in time, I wasn't even aware of this alternative understanding of the term.

In telco, when a remote node crashes at a client's site, I often only have access to a heavily restricted subset of logs, and the debugging communication loop via email can take days to understand "what happens". Because of that, I write defensive, strictly encapsulated code, and I think in terms of domain-specific states and objects that can be explicitly tracked from an external PoV.

Similarly, during game jams, "debuggable and maintainable" means to me that the code is modular enough that I can completely rip out and rewrite a core mechanic in the final 3 hours just because the game design suddenly changed.

My habit of writing code optimized for remote logs and sudden architectural shifts actually became my biggest enemy under the algorthimic interview (or 45-minute LeetCode) constraint. It makes the core algorithmic state less clear and hides algorithmic mistakes under layers of defensive "if" statements (where I would normally drop a debug log).

I am simply used to not trusting the inputs, whereas in algorithmic problems, the whole point is to exploit constraints that you need to be absolutely sure about.

So the "if" statements that usually increase "debuggability" in telco or during game jams are the exact opposite of the "debuggability" term used in algorithmic thinking.

Thanks for naming this issue so clearly - it is a very valid reality check.

This is so cool! I would love to hear what stavros ends up using it for – actually, in general I'd love to hear more about how folks are using voice recorders. It seems like now's really their moment as STT finally doesn't suck too bad and natural language can actually be turned into stuff.

I use it to talk to my LLM assistant! It's great, it's always there, always works, and it's very accurate, as you say.

I've been quite surprised, it even does Greek very well!

Maybe it's a silly question, but what are you using the assistant for? I love the idea of voice interfaces. Matt Webb's thing about distinguishing data and instructions by addressing "Diane" (https://interconnected.org/home/2025/03/20/diane) to work through a more substantive task by talking has some magic in it. And yet I've rarely gotten more use out of these things than, you know – toggling lights, setting timers. Plugging them into more meaningful integration seems really interesting.

It's a lot of tiny things, for example having all my info in one place when I go traveling, and being able to say "how do I get to the Airbnb from the airport". Yesterday I had the agent search for events in my city every week, find the ones I like and put them in a newsletter that it emails to me. I also had it look at my book history and ratings and recommend new books. Tracks my gym sessions, what restaurants I like and what I want to get next time, etc.

Nothing you can't live without, but it reduces the thousand little paper cuts of daily life by a lot.