So very recently there was an article about an SAAS provider baremetrics having to do a ton of due diligence for a buy that never happened. I own parts of a few properties that aren't publicly traded. Find my net worth(a real estimate without lawyers and months).
I am imagining it is a status thing. 90% of my worth is in commercial businesses with no liquidity. You can't verify that well without months and months of audits and accounting if my last sale is any gauge.
So I am high net worth. One issue is that 1) why would
I want people to know. 2) How won't this make me more of a target. The biggest issue since getting money is family/friends knowing. Why would I use this?
That reminds me of a version of an old joke about an old RAF or some other European forces pilot giving a talk. He starts describing a dog fight he had one time around 1914. "There were Fokkers to the right of me and Fokkers to the left of and one Fokker in front of me. The Fokker in front of me starts firing..." He stops as a grad student in the front stands up. "Sir. I hate to interrupt but Fokkers didn't have
a synchronization gear until 1915." "Yeah but these fuckers were Saulniers's." Not very factual, but kind of funny.
The idea looks interesting, but I don't feel like the examples show a strong enough improvement over just using multiple windows and some pdfs and a notebook.
The subtitle gives away the issue. They make biology easier to scale. The one and only one way to do that is to reduce complexity. They have some method that works in a restricted case. That is the benefit and deficit of using the platform. From working on flow cytometry software for >3 years I can say that has to be the case.
One really nice thing about older manuals is that they describe the entire system since releasing it was a hassle. I wish more web framework has as good of manuals instead of half the api being 'will document'.
So this is interesting. But it doesn't deal well with the idea that the speed of light may be variable across time. Not a particularly mainstream theory but the idea that the speed of light depends on time is kind of fun at least to consider.
I waited a few hours and didn't see any mention of the big one : https://wiki.osdev.org/Bare_Bones. In general osdev has a lot of good resources although the good threads get buried in the forum. Sortix came from there and the threads about trying to minimize the source dependency graph helped a lot in my first startup.
The more relevant link is https://keplerlounge.com/neural-computation/2020/01/16/compl.... Where the author discusses why this repos exists. This isn't a very good way to do AutoDiff and isn't exact(Look at standard reverse mode AutoDiff or Julia or Swift's implementations instead). The point of the repo is that the author believes it is a biologically plausible mechanism for neurons to be trained. Whether it is or isn't is the interesting question.
Imagine a doctor working at two hospitals who has to move his car from one location to the other mid-day between surgeries. Is it more fair for him to have a worse parking spot and take longer to get to the operating room because he didn't park there earlier? Or should spots be given based on need?
I think the argument is that this isn't to the benefit of other students, only to the benefit of other students that park improperly. Improperly parked cars decreases parking supply and puts more stress on students who park properly since they are unlikely to double park or park in a non marked location, whereas bad actors would be willing to deal with the low supply by parking even more improperly.
I think a lot of this is due to the fact that people don't understand statistics well. 30% confident is actually pretty confident if you know the full range of options. Most people with folk statistics tend to believe that anything less than 50% won't happen.
In general in WebDev and general software development there are lots of talks like these about how you should identify and overcome imposter syndrome. For some that is useful. For others, there was a fellow earlier who tried converting long(64 bit on Linux and 32 bit on Windows in C) to double and back. A double can loselessly hold a long smaller than 2^53 and a few other long's sparsely. One of the longs a double can hold is -2^63 which if he knew about doubles would not have surprised him. Meanwhile 2^63-1(the maximum value for a long in 2's complement when it is a 64 bit value) was not a double. My point being, be humble.
This is some article from a site for PIRG's. I am guessing from clicking around that is Public Interest Resource Groups? Either way most grad students want their work read and used. Most grad students want to be able to read and cite other works in theirs. Sci-hub at whatever address does more for students than any other source so far.
I feel like they attack from the wrong angle. Consider BareMetrics an SAAS. They aren't really just a CRUD, but are lumped in with CRM's. The easiest CRM is a spreadsheet and a mailer like SendMail.
This really feels like an article that is trying to say something important about a bigger issue. But the examples and the overall tone really diminish it. They allude a lot to things more important than them like Goodhart's law, mainly for cache, while they talk about their own poor quality idea(a bizarre imagined scenario far removed from, at least my, regular concerns). In the case of an AI show a concrete example involving a robot vacuum. Otherwise it feels like one of my junior's blowing smoke up my ass about hardware threads.