sanderjd

12 ms·

331.
by sanderjd 5mo ago
This does seem like an abstraction leak though.
332.
by sanderjd 5mo ago
I don't know who any of these people are. This seems very gossipy.
333.
by sanderjd 5mo ago
I generally agree with you (and disagree with the parent comment) that I think this seems like a useful technique. But just note that if someone really wants to break the contract, they can just use the "private" class. Just like
334.
by sanderjd 5mo ago
The problem I have with buy vs rent calculators is that the speculative questions often end up dominating the decision. How much will rent go up in the place you rent? Nobody knows. How much will the value of the property appreciate during
335.
by sanderjd 5mo ago
Yep, it sucks to be at a landlord's mercy to determine how nice the place you live remains over time. Things will wear out. If you own the place, yes, it's expensive to fix that stuff up. But if you don't own the place, you m
336.
by sanderjd 5mo ago
> Why use Claude then? I don't understand this question in response to the quote preceding it. He just explained why Claude was useful in this instance.
337.
by sanderjd 5mo ago
Yep, this is how AI has been impacting my experience at my job as well. For a given time and quality budget, we can now say "yes" to more projects. Often that means holding the time and quality constant and doing things we wouldn&
338.
by sanderjd 5mo ago
There but for the grace of god go we...
339.
by sanderjd 5mo ago
Are there any harnesses that are attempting to optimize for using local models like this? Unsurprisingly, my naive attempts to integrate with harnesses designed for frontier models have gone poorly. But it seems like a harness that understa
340.
by sanderjd 5mo ago
What is your startup?
341.
by sanderjd 5mo ago
Is this project public or have you written about it anywhere?
342.
by sanderjd 5mo ago
My favorite thing with Claude lately is it tells me things like: > I think this is a great thought, but I want to engage with it seriously rather than just agree, so for the sake of argument, ... I keep telling it to be blunt, that I c
343.
by sanderjd 5mo ago
Yeah this is the way. Thinking about how it will verify that it has done the right thing is the key. This been be set up at the AGENTS.md level. Very simple things like: "Write tests and make sure they pass." "Run lint after
344.
by sanderjd 5mo ago
Weird, this isn't my experience at all (mostly writing Python lately). Granted, it usually doesn't implement things exactly the way I want them to be implemented, and I iterate a lot on that. But I think it's been like a
345.
by sanderjd 5mo ago
But I don't relate to "it probably won't document its reasons"... I spend more time telling it that it has over-documented than under-documented. It always wants to include the specific thing it is fixing in the API docs
346.
by sanderjd 5mo ago
Yeah I work in pretty small chunks and do a lot of iteration on each chunk. This is harder when I'm reviewing other peoples' work. It has definitely gotten to the point where I spend more time reviewing than implementing and find
347.
by sanderjd 5mo ago
Yeah this is a good call out. What I've been doing so far is just hoping and praying that people naturally evolve their entire way of working without becoming demotivated or burned out. Seems like a bad strategy.
348.
by sanderjd 5mo ago
In my experience, those decisions generally (without additional prompting) end up in the PR descriptions, and then in commit history. Which is how it was with human written code as well... Or if it's more product level decisions than t
349.
by sanderjd 5mo ago
There is something to this, but to the concluding paragraph: I think these tools already are extremely good at helping us understand code, in addition to helping us generate it.
350.
by sanderjd 5mo ago
Yep, the first killer app for AI was, "hey, I'm having this problem and I'm not sure what's going on, can you help me figure it out?", and being told "sure!" instead of "this is a dupe, use the search
351.
by sanderjd 5mo ago
"I'm an agent person. I'm good at dealing with the agents! Can't you understand that?! What the hell is wrong with you people?! I'm good at dealing with the agents!"
352.
by sanderjd 5mo ago
Yep! But if I were doing my job as an accountant, I would not spend time doing any of that work using long hand division, I would use a spreadsheet. To me, it's a professionalism thing. If I'm doing a job, I should use the best to
353.
by sanderjd 5mo ago
High. My 2023 to early 2025 usage of AI was as "slight improvement to my existing editing and autocomplete capabilities". That was great and I loved it. But sometime over the last 12 months it has switched to "mostly using th
354.
by sanderjd 5mo ago
That's why I gave a range. I didn't say it is saving me 30 hours every week, I said 10 to 30 hours a week. So 30 is the max of the range, and I'd say the distribution is pretty heavily left-skewed. It really depends on what I
355.
by sanderjd 5mo ago
No, what is rewarded is "the code has been shown to conform to the given objective function" and especially "that objective function is a good representation of what we are trying to accomplish with this code".
356.
by sanderjd 5mo ago
I've been doing this for a couple decades. I don't wonder how people did it before AI, I did it for years and years before any of this existed... > What I found with LLM usage is that people will settle on the first one, decla
357.
by sanderjd 5mo ago
Yep!
358.
by sanderjd 5mo ago
I genuinely don't understand what you're talking about with this comment. Learn how to do what things properly? I've been writing software for two decades... I'm not primarily in a learning phase, I'm in a doing pha
359.
by sanderjd 5mo ago
I guess I just don't really agree that doing the tedious mechanical things is all that helpful for building the necessary mental model. I mean, I do think it was useful (indeed, necessary) for me to actually type out very similar lines
360.
by sanderjd 5mo ago
Agreed, it has always been the case. But I've never thought of it that way so explicitly. And I might argue that the important distinction is that the objective function is programmatically verifiable (which the word "requirements

More ›