Of course, if the system actually uses a resolution that accurate, it's quite likely you won't even be your own doppelganger. Retaining water would be enough to throw it off.
I'd wager that evolution and neural net learning has struck a pretty optimal balance between false positives and false negatives for this in the human brain.
Oh, and the human system definitely uses measures the researchers didn't take in this study; ever failed to recognize someone because they're not in the same context you usually see them? Heck, I once spent 2 hours on a train talking to someone I thought I sort of recognized. A day later I realized I'd been talking to a (former) CEO of one of the biggest companies in the country, but of course, I wasn't exactly used to seeing him outside TV or newspapers. Ah, well, at least he got an early insight into free software.
Because consoles don't usually render at it. Most games on current consoles render at 1600x900 or 1280x720, usually at 30fps and are then scaled to whatever screen you have. So setting a target of 1080p60 for a PC to be like a console is not accurate for most games. They have a few that are like that, but not many.
It isn't necessarily zero sum, since what they will own is a Mac laptop. While you can play some basic games on that, you'd need a second system to do any real gaming.
The logical argument of "Well just get a PC and use it for games and writing," is lost on them. I mean how could you NOT have a trendy Mac laptop? It is just unthinkable!
The commission is the only entity that can propose legislation. Usually, you do elect the people who can propose legislation.
The power of the actual elected body, the European Parliament, is still quite limited. They don't even have enough power to prevent their forced relocation from Brussles to Strassbourgh every month, rather being caught in a perpetual schoolyard bully 'stop hitting yourself' moment. They've managed to block legislation, what, once in history?
There are good and bad things about the EU, but democratic credibility isn't one of the good ones.
That is unlimited data at 2G speeds (around 100kbps usually), 2GB of data at 4G speeds. I love T-Mobile and have been a customer for quite some time but it isn't unlimited data in the sense many people would expect. On their base $50 plan you get as high a speed as the network can support for the first 2GB of data, then they throttle it down to a slower speed.
If you want unlimited (barring abuse, if you go too nuts they still might throttle) high speed data that is another $45/month.
Their base plan is the best plan going though. Really these days I think many people will find 2GB more than sufficient since WiFi is everywhere.
Yeah, I think it's quite common in most places for the same reasons you mention. And a human or an autonomous car with more situational awareness can deal with those situations by being extra alert and careful and keeping a pre-planned problem resolution at hand (ie, ensuring there's space behind/at the side to emergency break while passing).
But the Tesla just drives right into it, keeps a speed that might as well have been planned to make it invisible to the truck, while closing the distance to the car in front in a place where the road layout makes it look like it's intentionally cutting people from the merging road off from being able to take the exit. It's really bad behaviour.
I can't even say that it's good that it's collision avoidance managed to avoid the collision, because it that's how it works, there's a high chance that people will start to figure 'oh, look, I'm being cut off by a Tesla being a dick. No problem, it's fast enough to deal with me forcing myself into its lane.'.
In the near miss video the Tesla is engaged in behaviour that is so dangerous that it's illegal in many countries, as it's overtaking the truck in the outside lane. The legality might be a bit mitigated due to what seems to be a road merge right before, so the lane speeds might not have gotten sorted out, but considering the off-ramp or whatever it is that the truck is heading for, it's a traffic situation where exactly what happened is highly likely to happen. A situation where most human drivers would be very, very careful about exactly what that truck was doing if they intended to pass it. And where any real autonomous car should absolutely not be moving faster than the cars in the lanes to the left at anywhere near highway speeds.
And it works most of the time, the autopilot keeps the car on the road and avoids danger. Except for that 0.01% when it fails and you have to react as quickly as if you have been driving all this time.
The quality of a system is always measured in how well it handles exceptions. (Control question: Try to come up with a single example of a good system that handle exceptions badly. Hint: give up because such systems do not exist)
So a autopilot driving car will handle the normal case extremely well, but when something unexpected happens a human driver is much better capable of performing a sensible action.
Undersea cables are interesting beasts. When you look at them they are MASSIVE and so you figure there are a lot of pairs. Nope. 4-8 usually. All the rest is shielding and power. The big limiting factor size and cost wise is the amplifiers. You have to have a bunch of optical amplifiers in-line with the cable, and those have to be powered from the shore. Obviously each channel needs its own amplification so in the case of 6 pairs that's 12 amps. You then need a set of 12 amps periodically along the cable.Every few hundred km or so.
Hence, undersea cables are small in count when laid. Very different form land. If you hook up a building, fuck it you probably lay 144 fibers minimum because that's a small sized bundle. However for those long-haul undersea connection, it is just a few fibers per massive line.
If you are doing IT for an enterprise, get stats like this to go to management and show them why you need storage with snapshots and backups to alternate storage. Ya it costs to get a good setup, and it takes some IT time to administer, but all it takes is one of these and it has paid for itself.
We got hit with cryptolocker back in the day, the Dean opened it and it proceeded to go and encrypt the entire administration share he had access to. However we didn't pay shit, I went in to the management console, rolled back to an earlier snapshot, and we were good. Minimal disruption. Even had it somehow been able to blast the snapshots (users don't have write access to them so I can't see a way) we could have pulled data from tape that was at most a couple days old.
There's other reasons to do this too, of course, but this is a big one that is very visible these days, and so worth it.
Dunno if you've ever played with overclocking but the relation between speed increases past a chip['s normal limits and heat increases are not linear. So if you don't need/want the extra power, it is not a good idea to run your card harder than needed, particularly since those fans can get noisy when they spin up.
Support and stability are another reason. If a company rates a chip to a given level, they'll support/replace it there. Past that, maybe there's problems, maybe there's failures.
There can be a difference between what you normally design for and what something can actually do. Like my processor is spec'd to 140 watts TDP. What that means is Intel certifies that at standard operating frequency and voltage it will never dissipate more than that amount of heat, so if my thermal solution can handle that, I'm good. However the chip itself can survive more power, if there's a good enough solution. I can push it past that limit, if I want. If I do though, it is on me with regards to cooling and stability. If the chip messes up, they won't replace it.
What kind of money do you think those review sites make per review? It is going to be a real problem staying in business if every review is predicated on being able to purchase one (or more) of a piece of high end hardware that can cost in the real of $700 in this particular case. It might be a nice idea to think they'd do it out of the goodness of their hearts and just spend their own money to help people but that isn't how things work. They need to get paid and they all have to cope with the rise of adblockers dropping revenue. So ya, review samples are pretty important.
Also in some cases samples are sent prior to public launch, so reviewers have to to get their review ready for when the public can purchase the hardware, so they don't need to sit around waiting for a review or go in blind. These things take time to do right, getting them hardware early is the only way to make that happen.
If you have a solution where people can still get the hardware early enough time to meet release day reviews and can do it cheaply enough to be able to pay their people to work on reviews for a living, I'd love to hear it. However the reality is I don't think there's a way.
It was a valid lawsuit. I could see hating on someone if they were funding a long, drug out, suit with lots of delay tactics over nothing to try and force a settlement or bankrupt the other side. However the Hogan suit went to trial, and Hogan won in short order.
I don't see anything bad with someone funding a legitimate suit.
Advancements in car safety are great. When you look at the stats on car deaths despite number of miles drive per year going up, death rate just keeps dropping. It isn't because people are better drivers, but because we've managed to build in so many safety features in to cars. I love it.
I'd never heard of that 911 feature until your post. Something like that is wonderful because it means even if you are completely incapacitated EMS is summoned quickly.
You can't go home again, unless you set $HOME.