dekhn

32 ms·

631.
by dekhn 1y ago
Can you not post these diatribes here? They are incoherent and irrelevant to this site.
632.
by dekhn 1y ago
It seems improbable from an economic perspective, given the challenges associated with launching the hardware, maintaining and operating the hardware, heat dissipation, and the lack of use cases for which space data centers are uniquely poi
633.
by dekhn 1y ago
no. for internal training most work is done on TPUs, which have been explicitly designed for high performance training.
634.
by dekhn 1y ago
Larry's original goal for Google was always to be a revenue vehicle to reach AGI although I don't think Sundar is interested in anything except revenue/profit. Note also that many of Google's previous attempts with LLM g
635.
by dekhn 1y ago
Lack of foresight in AI? They own one of the world's largest GPU/TPU fleets. They have some of the most advanced research and a path to deploy products. What exactly else should they have foreseen?
636.
by dekhn 1y ago
For DNA? Not sure. https://pmc.ncbi.nlm.nih.gov/articles/PMC3197634/ claims it's reached parity on protein sequences. I'd much rather have a slower, but more accurate searcher, or one that was easier t
637.
by dekhn 1y ago
I've also worked with probabilistic diff- like tree-based, but tolerant of parsing errors.
638.
by dekhn 1y ago
The creator of the Myers algorithm is Gene Myers. He also helped create the BLAST algorithm, one of the fastest and most important DNA and protein search algorithms, and also implemented most of the original human genome assembly done by C
639.
by dekhn 1y ago
Also: all the Alton Brown Good Eats books, and On Food and Cooking (McGee). Much of cooking is physics and chemistry. "The science of YUM"
640.
by dekhn 1y ago
an enormous amount of software was developed on the PDP-10 and PDP-11 and later VAX systems that could not have been done on microcomputers in the day. You can't just compare raw RAM and clock rates, the PDPs were set up for multi-use
641.
by dekhn 1y ago
Likely any early implementation of ML would have been on a mainframe or minicomputer, not a 6502. A mainframe/minicomputer would have had oodles of storage (both durable and RAM), as well as a compiler for a high level language (which
642.
by dekhn 1y ago
Larry Summers: https://www.nytimes.com/2005/01/18/us/harvard-chief-defends-... He clarified his intent: https://www.harvard.edu/president/news-speeches-summers/2005...
643.
by dekhn 1y ago
MD was never really a viable way to do structure prediction, so it didn't become obsolete with AlphaFold. Instead, MD is more useful for studying the physical process of protein folding (before the protein folds to its final structure
644.
by dekhn 1y ago
Probably better to say that we contain multitudes of probabilistic graphs that resemble state machines, but those graphs do not make up our entire system. Further, those graphs interact constantly with stochastic systems (the environment,
645.
by dekhn 1y ago
That's surprising given how badly restoration worked (much more like tape than drives).
646.
by dekhn 1y ago
No, I don't really argue about individual studies with individuals on the internet. What I'm describing is the current consensus opinion of the larger medical and research community.
647.
by dekhn 1y ago
Title is wrong- the linked article does not claim the original study was faked, just that it failed to properly correct for bias in the data.
648.
by dekhn 1y ago
there is more scientific doubt about the true effectiveness of COVID vaccines than you think.
649.
by dekhn 1y ago
It's mostly used when the acronym is pronounced like a word (Nasa). They don't write "Bbc" because it's pronounced "bee bee cee" or "beeb"
650.
by dekhn 1y ago
Ultima III/IV
651.
by dekhn 1y ago
Here's another book from my childhood that I never forgot (before reading the solution in Wikipedia, consider buying the book used and trying it yourself): https://en.wikipedia.org/wiki/Masquerade_(book) We used t
652.
by dekhn 1y ago
Ah, I was going to post asking if anybody else read all the pages looking for unreferenced endings.
653.
by dekhn 1y ago
Another issue some SGI (O2) had: they didn't clear the entire video/memory buffer between logins. So, occasionally, when you logged in on the console, you'd see random images from the previous user on the screen. Apparently
654.
by dekhn 1y ago
Same as Superman I -Lex Luthor steals a nuke to hit the San Andreas fault and turn Luthor's real estate into beachfront property.
655.
by dekhn 1y ago
I grew up near a town called "Moodus" in Connecticut which constantly made noises and had small quakes. But it didn't prepare me for the few small quakes I experienced in the bay area (typically a bunch of car alarms go off a
656.
by dekhn 1y ago
triboluminescence https://www.nytimes.com/2007/06/19/science/19winto.html
657.
by dekhn 1y ago
Beowulf style clusters went on to dominate supercomputing.
658.
by dekhn 1y ago
Parthenogenesis is not uncommon in animals: https://en.wikipedia.org/wiki/List_of_taxa_that_use_partheno... (I am mostly quibbling with "rare few animals" but I can't really say much about the relative p
659.
by dekhn 1y ago
We are truly spoiled by all the improvements that went into local filesystems that are lacking in network filesystems. So much of our perception of "computer is fast" is really just write-caching, read-caching, read-ahead.
660.
by dekhn 1y ago
I have been working with 6502 the past few days. Apple IIe hires mode was the bane of my childhood existence, but these days it's not so bad- an LLM can easily write code (C or assembly) to do high performance drawing. Unfortunately,

More ›