dekhn

34 ms·

901.
by dekhn 2y ago
Yes, a part of the government can try to hurt the government, that is true.
902.
by dekhn 2y ago
Without making any statement on whether I believe DOGE is being seditious, it's not hard to interpret their underlying intent as being wholly compatible with the definition you pasted. DOGE intends to hurt the government in ways that
903.
by dekhn 2y ago
I don't believe your statement that you ask for, and successfully receive, tenant admin rights from large government departments. DOGE employees aren't simply doing their job. They are actively subverting the government to fatall
904.
by dekhn 2y ago
Note in this case, the fault was not the paper authors- the editor in chief retracted the paper because the paper said the peer reviewers asked for irrelevant citations to be added, and the EIC concluded "peer review was compromised&qu
905.
by dekhn 2y ago
I think CS is math, not science. Computer engineering is science (using lots of math).
906.
by dekhn 2y ago
Appreciate the sarcasm, but... it's really 3 billion years of evolution, with astronomical levels of actual entities living and dying in a dynamic world environment. Chemical reactions happening in nanoseconds. Polymers have extraord
907.
by dekhn 2y ago
I invested a great deal of effort over 30+ years to learn biology, which I started to love in high school when a teacher introduced us to molecular biology. Over time I've come to appreciate that biology is a huge field and people w
908.
by dekhn 2y ago
I use this as one of my interview questions. Basically, identify the pattern, convert to/from serialized integers, do linear extrapolation based on a few observations over the years (of the highest seen plate number). The candidate c
909.
by dekhn 2y ago
It was the trifecta of high holy days (Passover, Easter, 4/20).
910.
by dekhn 2y ago
mitochondria work by using proteins. Many mitochondrial diseases are due to misregulation of proteins. few molecular pathways are untouched by a protein at some point. Research into fundamental properties of proteins is important because
911.
by dekhn 2y ago
slippery slope isn't a fallacy, and we're discussing matters that can't be directly analyzed using purely logical arguments.
912.
by dekhn 2y ago
https://www.nature.com/articles/s41567-025-02865-1 (paywalled) and https://www.researchgate.net/publication/384256048_Shape-Rec... I don't know how the popular mechanics author came to the co
913.
by dekhn 2y ago
He's certainly not the only one, many people who live in rural or semi-urban areas will occasionally eat roadkill. It's considered safe if you follow a collection of precautions. I think he's an idiot but I don't think
914.
by dekhn 2y ago
yes, that was one of the externally shared narratives. The other part is that google didn't want to be beholden to nvidia GPUs, since they have an associated profit margin that is higher than TPUs, as well as resource constraints (tot
915.
by dekhn 2y ago
I think the choice they made, combined with some great software and hardware engineering, allows them to continue to innovate at the highest level of ML research regardless of their specific choice within a reasonable dollar and complexity
916.
by dekhn 2y ago
I have not seen evidence that the methodology of science is applied routinely in sociological research.
917.
by dekhn 2y ago
The route was also adjusted to be far less practical as a compromise to get the central valley politicians to approve it. I really wish they had done a straight shot between Burbank and San Jose but even then it would have been extremely e
918.
by dekhn 2y ago
Yes, as practiced, as well as in principle, sociology is less scientific than physics. Some systems- human social behavior being one of them- are simply not amenable to modern science because the systems have far too many uncontrolled varia
919.
by dekhn 2y ago
A 3d torus is a tradeoff in terms of wiring complexity/cost and performance. When node counts get high you can't really have a pair of wires between all pairs of nodes, so if you don't use a torus you usually need a stack of
920.
by dekhn 2y ago
As an aside, Google used CPU-based machine learning (using enormous numbers of CPUs) for a long time before custom ASICS or tensorflow even existed. The big ones were SmartASS (ads serving) and Sibyl (everything else serving). There was an
921.
by dekhn 2y ago
Google shouldn't do that comparison. When I worked there I strongly emphasized to the TPU leadership to not compare their systems to supercomputers- not only were the comparisons misleading, Google absolutely does not want supercomput
922.
by dekhn 2y ago
this is basically George Church's MO. I respect him for his early work in sequencing, and he has some crazy/great ideas, but he also oversells everything to the press, which eats it up and spits out articles with the narrative &q
923.
by dekhn 2y ago
Having opinions is one thing; disrupting a major meeting with an unrelated complaint about your employer (and generating national news) is another.
924.
by dekhn 2y ago
Much of the course topics look great. The instructor looks like a jerk.
925.
by dekhn 2y ago
Schools don't want people who min-max for college acceptance. They want people who are genuinely excited to go to that school. I don't think his essay was necessarily a deal-breaker if you read it as a "I learned from experi
926.
by dekhn 2y ago
I am so glad I don't have to take that class.
927.
by dekhn 2y ago
I consider UC's statement of diversity ( https://www.nytimes.com/2023/09/08/us/ucla-dei-statement.htm... and then 6 months later https://www.nytimes.com/2025/03/20/us&#
928.
by dekhn 2y ago
I think it comes from a misread of the text in the gates interview linked in the comment: "r. We were moving ahead very rapidly: BASIC, FORTRAN, LISP, PDP-10 machine language, digging out the operating system listings from the trash an
929.
by dekhn 2y ago
Endowments are typically restricted funds (imposed by the fund provider) and can't be used (unless the restrictions are removed) to be used for general operating budget. Harvard generally uses the interest on the fund principal to pay
930.
by dekhn 2y ago
I also found a DOE grant, about $800K. I think this is the full list, NIH looks like a subset of overall HHS funding, and NSF is the actual single largest (around $2.5M) https://ncsesdata.nsf.gov/profiles/site?method=re

More ›