time "spent" in the database depends on the speed of DB drivers too, which depends on the language. Our stuff is Java and average query is ~2ms round trip.
Python, Ruby and friends are bad for page speed metrics. REST response time of ~200ms vs ~10ms for our java backends
And working less hours is also skewed by the rise of millions of gig jobs that don't have set hours.
The top 25% of the population is doing great by any measure. But when you discard the top 25%, the rest doesn't look so good. Income inequality in US is higher than almost every other industrialized nation. As high as it was in the 30's when massive social unrest gave rise to unions and worker protections.
When you factor in student loan debt and minimum wage being degraded by inflation and gig jobs, the bottom 50% are poorer than ever. As of this year, student loans have eclipsed credit card and auto loans as the second largest form of consumer debt behind mortgages
It might be a good thing for me at this point. Started out on a promising new project. Years later its been designed by committee into a sprawling unfocused mess. Maybe getting the boot would spur me to go back through the interview gauntlet
Unless the site uses certificate pinning its possible to do a downgrade attack that forces browser off of HTTPS. The extension HTTPS Everywhere is a stopgap against this
To blue eyes again, you could make a facial recognition system that just looked for 2 blue iris shapes and probably score 90%+ accuracy on most datasets (of exclusively blue eyed people and non-human objects) because blue is so rare in animal eyes, and so rare in nature. With such a strong face signal it may be impossible to match accuracy for non-blue-eyes humans without purposely making match quality for blue eyes worse.
We may be seeing a similar but weaker bias for darker skin. Light skin is fairly rare among mammals. The only solution to eliminating recognition accuracy bias may be to nerf accuracy for people with light skin. There have been many reports of facial recognition erroneously classifying dark skinned humans as animals. If part of model accuracy classifies lighter skin animal = more likely human, its going to be very hard to remove that bias because its empirically true. If the classifier is unsure if a shape is a human or animal face, but its very lightly colored, it may make the model statistically more accurate to report higher confidence that a lighter colored face is more likely to be human.
this may be more of a "most animals are brown" than a "brown people look more like animals" problem with the models. A possible path to fixing the bias is to crop all faces detected, disregarding model confidence of whether the face is human
I'm going to be controversial here, and say that its possible that light colored skin is easier to identify from background, giving higher confidence to "is face" detection with lower false positives.
Brown is, next to green, the most common color in nature. Lighter color also has a greater contrast with the background when using camera flash or other subject lighting.
To be really blunt, I think white faces are easier to see. A common example in nature, white tailed deer. When tail is raised, their white fur is a clear danger signal because it stands out from the background. Like most mammals, the rest of their body is a shade of brown.
I don't think its racist to suggest that light skin, just like blue eyes, is a conspicuous highly visible signal suggesting a human face.
I find it hard to believe that such a large number of facial recognition models have an easier time identifying white people after years of publicity solely because of biased training data.
The ultimate example is probably blue eyes. Almost no other animal has blue eyes. Blue is extremely rare in nature. I want to see a study on facial recognition, blue eyes vs other colors. I bet money you will find that models excel at recognizing human faces with blue eyes above all other factors
> Tech companies have made clear that they don’t like the idea of blocking apps without a more organized policy process, and have suggested that they see this as a First Amendment issue, said Adam Segal, a cybersecurity expert at the Council on Foreign Relations.
Yet they're both perfectly fine with blocking apps for arbitrary reasons citing "app store policy".
Lets be blunt here, whether you agree with the Tiktok issue or not, for Apple and Google this is purely a money issue. They are probably afraid this will increase support for third party app stores among the masses
As far as "studio monitors", you should know that a quality set of over ear headphones has better sound reproduction than any speaker. I used to be involved in the pro audio scene, until I realized this.
Its impossible to eliminate echo and stereo bleed with speakers. If you try to EQ a room using mics you will see this clearly. You might get close building an anaerobic chamber, but headphones are still better.
If you're on a budget don't waste money on expensive speakers, just use headphones
India and China aren't exactly friends, I think its pretty reasonable to limit business ties of any kind. A sovereign nation has the right to ban any import, are you disputing that?
By most anecdotal accounts GS-441524 is better and possibly safer. And its readily available on the black market for treatment of cat FIP because apparently its much easier to make than Remdesivir.
So yes, probably the only reason Remdesivir exists is patents
I never "bought" TDD. Just like "microservices" it adds a ton of complexity and cost for something you probably don't need.
The TDD cult doesn't consider that its the last thing you should try, test everything. If you need high reliability you should move to a safer language first. Dynamic typing to static. Unsafe memory to safe, like Rust or Go. Turn on a bunch of linters too. Move to a language that doesn't allow nulls.
Once you've got a bunch of linting and a static language that doesn't allow memory corruption, do you really need to test everything? Probably not. The linux kernel is a good example as usual. Virtually no tests and its the backbone of the internet
anything with an orbit calculated to take more than a decade to decay without station keeping is banned. These calculations are routine so the requirement is only a burden for mission cost (needing more satellites for coverage). Exceptions for scientific missions and satellites used for extra-earth communications
Useful but not necessary. Anything pointing at the Earth, nearly all satellites, work better closer. The issues of line of sight and coverage are all about money, and the cost of small satellite launch is dropping rapidly.
There are some satellites like telescopes that need to be far from the earth. But the vast majority could be banned from orbits beyond a distance where they won't decay in a reasonable amount of time, with the only downside being cost
I think there needs to be rules about orbital distance. Low earth orbit satellites like Starlink aren't risky because their orbit decays rapidly without constant boosts. Its the perfect failsafe.
The only reason to use higher orbits is to save money on satellites and complex communication systems. Orbits that allow space junk to hang around for centuries should be banned.
Python, Ruby and friends are bad for page speed metrics. REST response time of ~200ms vs ~10ms for our java backends