mpalmer

29 ms·

601.
by mpalmer 2y ago
If I reviewed PRs like I comment on HN I'd get fired. Know your audience! Seems like you think PRs are the only place where criticism happens.
602.
by mpalmer 2y ago
Why is criticism bucketed with cynicism? I led with my appreciation of the good things in the post. When the day comes that I post something of mine on HN, I will be tremendously disappointed if all of the comments are the textual equivalen
603.
by mpalmer 2y ago
If you get anything as succinct and focused as what I (genuinely) wrote myself, I'll gladly take the criticism!
604.
by mpalmer 2y ago
Look, I love the sentiment, and the illustrations are charming. Unfortunately, the writing. It's...stilted. It's presented as a letter/email, but it reads as though the author wants you to hear someone with good comedic timin
605.
by mpalmer 2y ago
Underutilized, or over-prompted for the layperson?
606.
by mpalmer 2y ago
Wow, what apps have you made so I know never to use them?
607.
by mpalmer 2y ago
> millions of years more technologically advanced than us > should be leaving telltale signatures across the sky that we'd likely have detected by now I'm not sure the second follows from the first. What if they're hidi
608.
by mpalmer 2y ago
Why waste everyone's time with this? We can generate facile LLM responses ourselves.
609.
by mpalmer 2y ago
Quite a pronouncement to make with no supporting detail. Reads like troll bait if I'm being honest. But I'll respond in kind: Yes, it is. Tough is relative. There is no "we" to be coddled, only your flawed perception of
610.
by mpalmer 2y ago
FOIA requests can be made against formerly-classified information. But it's beside the point; any(?) non-classified information/communiques in government are subject to FOIA. Plenty of non-classified info in that chat and the ones
611.
by mpalmer 2y ago
Literally the only thing I've been addressing is the proper usage of the word bias, there is nothing implied, hidden or hinted at. You seem determined to make the definition of the word serve some AI-related concern.
612.
by mpalmer 2y ago
Again, the word "bias" and its definition exists outside the comparatively narrow concern of training LLMs.
613.
by mpalmer 2y ago
The definition of the word has no responsibility to your opinion of it as an epistemology. Also, you're just complaining about the difficulty of determining what is true. That's a separate problem, isn't it?
614.
by mpalmer 2y ago
The baseline is a neutral stance on orange. The option isn't biased, a choice isn't biased. The chooser is.
615.
by mpalmer 2y ago
Bias implies an offset from something. It's relative. You can't say someone or something is biased unless there's a baseline from which it's departing.
616.
by mpalmer 2y ago
"Humbly"? The author has full control over the title, and in addition to being bait, the title is not humble at all. Not a case against, but the case against.
617.
by mpalmer 2y ago
"other people's bad work makes it pointless for me to do good work"
618.
by mpalmer 2y ago
Here's my left-field pitch since everyone's shopping the A.IDEs - don't use them. The polish of Cursor et al is a component of its dishonesty. You should keep the nature of the tool exposed to you at all times. I say you shou
619.
by mpalmer 2y ago
Yes, strange to have so much and still feel like you'e lost something.
620.
by mpalmer 2y ago
Not all statements are headlines.
621.
by mpalmer 2y ago
I have! A more conventional story, but it too bears basically no resemblance to this ad whatsoever.
622.
by mpalmer 2y ago
And Apple could lose a billion a year for decades if it wanted to.
623.
by mpalmer 2y ago
This doesn't look like anything Lynch would have made. Quick cuts, tons of effects shots, straightforward narrative...
624.
by mpalmer 2y ago
You should choose what is right for you and accept that others won't like it, like anything else in life
625.
by mpalmer 2y ago
What does it matter what nonsense we in the US do? The Italian law is well-made and justly-carried out, or it isn't. Personally, I am happy to condemn my own government for its many failures.
626.
by mpalmer 2y ago
I don't think that's the problem. It's time and effort; junior devs (and more broadly, humans) always want to see how little they can spend; here they incorrectly believe they can lean on LLMs and get a guaranteed increase in
627.
by mpalmer 2y ago
> If, by spending enough tokens, you end up with a working feature, then this is a valid method of doing the work. It's just several really big ifs: - Does "no errors" == working feature? Only if the tests are good, and ev
628.
by mpalmer 2y ago
I wouldn't let junior devs anywhere near "agentic" tools like Aider / the newer Copilot stuff. These things just let you turn off your brain and spend hundreds of thousands of tokens just rewriting entire features until
629.
by mpalmer 2y ago
AI is fantastic for doing stuff you are already qualified to do, but faster. It's great for prompting you with what to do next, and offering different ways to think about the problem in front of you. But you have to be able to describe
630.
by mpalmer 2y ago
Easy to believe our weak privacy laws are part of the reason we get tech features first. Huzzah...

More ›