yobbo

26 ms·

391.
by yobbo 4y ago
If nothing else, US would do its best to prevent F-35s falling into Russian hands. In that way, F-35s grouped in bases close to the border is an effective trip-wire.
392.
by yobbo 4y ago
There were rumours of an Isreali F-35 targeted and damaged (but survived) by some old Russian AA system, maybe BUK? This was based on observed events and movements. Of course, we would never know, which is correct and appropriate. The topic
393.
by yobbo 4y ago
The Eurofighter is higher specced than the F-35 on a number points - speed, manoeuvrability, capacity, and possibly endurance - but this is not really part of the debate. The question is how they are going to be used in practice, with every
394.
by yobbo 4y ago
Somewhere on Hinton's webpage, there's a series of mp4 lectures. They introduce general neural networks, and then lead into restricted boltzman machines, which he worked on. Probably one of the best introductions to RBMs.
395.
by yobbo 4y ago
Debian now works fine without systemd. The alternatives are sysvinit, openrc and runit. I've removed systemd from many working installations. The problem is rather the many random dependencies. (Block systemd in a preference file with
396.
by yobbo 4y ago
Mikrotik makes wifi radio-links where the radio is integrated into the dish, connected by cat5 for example. The right solution to avoid lossy antenna cables/connectors where power is limited to begin with. For long distances, you will
397.
by yobbo 4y ago
This is not a benchmark of some model on cifar-10, it's a benchmark of the training algorithm. But, model size and complexity also matters. MLP with backprop gets about 63% on cifar-10, for various reasons. So achieving 59% accuracy me
398.
by yobbo 4y ago
Yes, but as a counter-point; how do you determine if someone is cosplaying? You seem to imply that there is some skill that makes startups a deterministic job, but it isn't. Everyone is cosplaying until they meet some challenge for whi
399.
by yobbo 4y ago
Ridiculous but "Panasonic", "Lexus", "Sony", "Pentax" and so on are made-up names intended to sound meaningful in western ears. Maybe they should "lean into" Japanese words like like how Den
400.
by yobbo 4y ago
> Money is the goal, code is a tool to get the money. This is an obvious and empty statement if it's not used in specific cases. Unless you have something new to add in the "how to get money debate" the distinction between
401.
by yobbo 4y ago
(It's not counting input tokens.)
402.
by yobbo 4y ago
20 megapixels is 5000x4000 which is unpractical, and always scaled down on the web. The phone lens aperture might be too small to capture meaningful details beyond a certain distant from the camera, so 20 megapixels is probably physically o
403.
by yobbo 4y ago
> Compare making it freely available to anyone The original idea of a patent is that it must reveal the "secrets" behind an invention, in return for a "temporary monopoly". It's supposed to help diffusion of tech
404.
by yobbo 4y ago
One could make an argument by comparing the effect on productivity by either replacing the worker or the machine. When workers are readily available (and machines are designed to make workers replaceable), it doesn't matter what a time
405.
by yobbo 4y ago
One can argue that neither execs or workers deserve more, but worker's pay is mostly determined by supply and demand, while exec's pay is a governance issue that's the responsibility of the owners.
406.
by yobbo 4y ago
Try using, and servicing, an old industrial lathe or anything else in a mechanical workshop and see how quickly you can get up to speed. Repeatedly achieving meaningful results, and keeping the workshop functional and clean, was someone&#x2
407.
by yobbo 4y ago
How would you determine if increased productivity is due to "labour or capital"? Are people themselves more productive, or are the machines just better? Everywhere I have seen, machines are doing almost all work.
408.
by yobbo 4y ago
The Rust-test implements the traits Add, Sub, Mul by value. This makes the few references less important in the total test. The ergonomics argument is motivated by using these traits. Otherwise, references would have had the same ergonomics
409.
by yobbo 4y ago
Same problem applies to Uber and other services though, although they take a higher "cut".
410.
by yobbo 4y ago
The phone numbers could be the public identities included in the rating signatures, (ex: "555-12345, 5 stars, <$10 signed by paypal>" signed by me.) Drivers are incentivised to stick to one phone number, and to publish these
411.
by yobbo 4y ago
I think I agree with the article in that the actual problem is to make drivers "accountable" with something that works like a rating. To achieve that without centralisation requires some variant of crypto signatures. To make them
412.
by yobbo 4y ago
It shouldn't be hard. Reputation could be achieved with some form crypto signatures and identity based maybe on phone numbers, with traceable identities. Blockchain not needed or helpful. (A rating signature must probably include a pay
413.
by yobbo 4y ago
> a belief system that the tribe's preferred framework is always the right tool for the job. It's not really about "the right tool for the job". It's about protecting and promoting the tribe's investment.
414.
by yobbo 4y ago
So it has a steep learning curve and it's not the "best" choice for people's first couple of projects. But, before you can make anything meaningful with a language, you have to first make many toys. That's the rati
415.
by yobbo 4y ago
My thoughts are that power plants don't stop completely in blackouts, they disconnect, power down, and idle. The reason they need to disconnect (in chain reactions) is that the load is too high for them to maintain frequency/kee
416.
by yobbo 4y ago
https://data.oecd.org/chart/6UQi It doesn't look as bad as I thought, but I suspect some national statistics are using evermore generous definitions of "participation". For example, people in higher educ
417.
by yobbo 4y ago
"Did not respond" is a legitimate result of an experiment and should be included when making statistics. Increasing sample size doesn't change this. The question itself seems confused.
418.
by yobbo 4y ago
> But this is ignoring the fact that automation never replaced workforce as such. But it actually did. "Workforce participation", which might be the closest metric of what we actually mean here, keeps declining. In addition to
419.
by yobbo 4y ago
10-20 years of experience in software engineering means around net €30k in Sweden, even in Stockholm. Tax and social security fees are the highest you will find, which means the cost of this might be around €70k-90k to the employer.
420.
by yobbo 4y ago
Not really. Any business can now use tools that make jobs redundant. It is having its effect everywhere. Anyone whose job is to copy information between mails, spreadsheets, and some webapp will eventually be replaced. There are more "

More ›