LPisGood

10 ms·

151.
by LPisGood 3mo ago
Not everything needs to “scale.” I think needing scale is relatively rare, actually.
152.
by LPisGood 3mo ago
I’ve heard of “blue suiters” for air force brass, but never blue badgers. Anyways, isn’t NSA one of the largest employers of mathematicians in the world? Surely they’re doing something useful.
153.
by LPisGood 3mo ago
Note that 200 dollars of value is different than 200 dollars of profit.
154.
by LPisGood 4mo ago
This is in contrast to American models which receive _ridiculous_ levels of private subsidy.
155.
by LPisGood 4mo ago
Probably 401 (k) plans for the most part.
156.
by LPisGood 4mo ago
How is it different than if I buy a TV and sell it on Ebay and use the funds to gamble?
157.
by LPisGood 4mo ago
It is, and I know someone who found that out the hard way during the Super Bowl.
158.
by LPisGood 4mo ago
> a Valedictorian, NHS member, JV or Varsity sports team member in HS, getting a 2100/1500+ on the SAT, and taking 6-7 APs are now table stakes This is very true in my experience, except I subbed out Valedictorian with multiple vars
159.
by LPisGood 4mo ago
You can. IQ scores are not stable throughout one’s life.
160.
by LPisGood 4mo ago
Their pre-screen test was awful. Brain teasers, including some infamous ones like “if you have two coins that make 15 cents and one is not a nickel, how is this possible,” and moderately involved programming questions like “parse phone numb
161.
by LPisGood 4mo ago
The point of this post isn’t that the “reasoning” phase of LLM thinking isn’t the same as what humans consider reasoning; it’s that Anthropic is intentionally hiding Claude’s “reasoning output” to make the model harder to distill.
162.
by LPisGood 4mo ago
People talk about this a lot. What I have never seen is a discussion of methods they might employ to degrade the models. Let’s say I’m a bad faith LLM operator, and I want to degrade my model so the next release looks better and people want
163.
by LPisGood 4mo ago
Don’t those uniformed maintainers get reassigned to other military jobs or are they allowed to work as a contractor while being active military?
164.
by LPisGood 4mo ago
OpenAI has notoriously terrible engineering. Look no further than the numerous Reddit threads and YouTube videos about people trying to give them money for their services and being denied. Users routinely have to try a number of credit card
165.
by LPisGood 4mo ago
My very small town growing up had sewing machines and they eventually even got a 3D printer. In high school I sewed a heart shaped pillow for a valentines day present; the library provided a bin of free fabric/stuffing as well as the m
166.
by LPisGood 4mo ago
I’ve tried this (not on HN) but usually it just results in downvotes or mockery.
167.
by LPisGood 4mo ago
That person doesn’t sound like an A at all in my opinion.
168.
by LPisGood 4mo ago
This is an interesting perspective(and a great roadmap for juniors to try to improve), but I think for the most part the thesis is wrong, at least in my experience. Companies do not hire juniors as some long tern play to develop them into g
169.
by LPisGood 4mo ago
Re: source code producing different binaries: things like ASLR, stack canaries, optimization levels, linking, etc all lead to different binaries.
170.
by LPisGood 4mo ago
That decision had a lot of detractors at the time, but it ultimately seems to have been correct.
171.
by LPisGood 4mo ago
To me it is almost tautological that what I said is true. If software can’t solve a knowledge work problem, that means the software isn’t good enough and needs improved. I also question your claim that “everyone hates software engineers,” b
172.
by LPisGood 4mo ago
I recommend everyone in this thread call OpenAI and talk that support agent. I had some issue and tried to trip it up, spoke naturally and ambiguously, and it actually did a pretty good job.
173.
by LPisGood 4mo ago
Exactly this, software engineering is “intelligence complete.” If you can solve it, you can solve all knowledge work.
174.
by LPisGood 4mo ago
Software has, with precious few exceptions, almost always been held together with duct tape and a prayer.
175.
by LPisGood 4mo ago
Isn’t that a black mirror episode? Anyways, much like the Matrix, using humans for energy is insanely inefficient.
176.
by LPisGood 4mo ago
Looks like 25 billion this time around.
177.
by LPisGood 4mo ago
I have some thoughts on that. It’s possible that the current administration’s lack of oversight has artificially propped up the stock market. If the next president decides to return to a more traditional rules based structure, the market wi
178.
by LPisGood 4mo ago
I believe my development tasks are some of the most complex and sophisticated out there (high performance mathematical optimization and modeling), and to me Opus 4.6 is pretty good. Which sort of development tasks do you believe are still o
179.
by LPisGood 4mo ago
Do we really care about this gap? If open models are 6 months to a year behind frontier models, does it really matter that much?
180.
by LPisGood 4mo ago
The concerns here, like most concerns about privacy, are hyperbolic hypothetical hypochondria, until they’re not.

More ›