You do realize that that cost is heavily subsidized by the government?

No, he doesn't realize that because it isn't true. USPS does not receive any tax subsidy. It is currently running an accounting deficit, but only because it's being required by law to pre-fund health and retirement benefits for the next 75 years in the span of a decade. If USPS wasn't being required to fund retirement for employees who haven't been born yet, it would be in fine financial shape.

But the problems with our health spending are not primarily in the public sector. Those other countries that have more efficient healthcare than we do have more of their healthcare run by the government, and there's a fairly strong correlation between cost effectiveness and government control. Within the US, the the government is generally more cost effective than the private sector. Within the government sector, the most efficient provider is the VA, which runs its own hospitals rather than just being a glorified insurance company. There's every reason to think that our healthcare system would be improved by turning more of it over to the government.

Don't neglect the cost of advertising, either. Paying for ads is a non-labor expense, and it can easily make or break a product..

In any case, complaining about marketing costs is often silly. An engineering team is basically a tool to convert money into new products. To stay in business, it has to be connected to another group that converts the new products back into money, which means some kind of marketing. You need both sides to pull their weight for the organization to thrive in the long term. As long as the marketing people are doing a good job of bringing in the money and aren't making promises the engineers can't keep, it shouldn't be a big deal to the engineers exactly how they do it.

It's actually a better comic than you might expect.

You say tomayto, I say tomahto. What else is money laundering but trying to keep your money out of the government's control?

The trick is that the shutter isn't doing the work; the flash is. It's possible to make very short flash pulses; I think you can make them even shorter than the 1/50,000 second mentioned in the article. As long as most of the light for the photograph comes from the brief but intense flash, the ability to freeze action depends on the flash speed rather than the shutter speed. You actually need to make sure the shutter speed is slow enough that the shutter is guaranteed to be all the way open when the flash triggers (X-sync speed or slower), or only the area behind the open part of the shutter will be exposed. Controlling things using the flash also guarantees that the multiple cameras used for 3D photography will all be taking their pictures at exactly the same instant.

Also note that the limitation you're talking about only applies to focal plane shutters (i.e. those right in front of the film or sensor). It's also possible to use a central shutter that's located right next to the iris of the lens. Central shutters open and close like the lens aperture, but block the lens completely when they're closed. Like the lens aperture, they block light to all parts of the focal plane more or less equally as they open and close, so they don't induce any of the motion effects that focal plane shutters do. Central shutters have their own problems- it's hard to make them work for very short shutter speeds, and they have limited efficiency when you use them that way because they're only completely open for part of the time- but they do eliminate focal plane shutter artifacts and allow you to flash sync at any available shutter speed.

You tell me another field that comes even close.

Easy: Economics. You have similar, if not greater, problems conducting controlled experiments, especially in macroeconomics, and there's even more money and politics involved. Economics winds up being closer to theology than it is to science, even though it's something that ought to be amenable to the scientific method.

If you find a journal that is reputable and like it, then "sign it".

Something similar is already formalized in academic publishing. When an author trusts an individual article, he'll cite it as a reference in his own articles. Articles that are important can be cited hundreds or thousands of times, while trivial ones may never be cited at all. If you take all the articles in a journal and see how many times they've been cited on average*, it gives you a good idea of consensus opinion of the quality of the journal. This is the basis of measures like the Impact Factor.

*You may wish to use some method of averaging other than taking the arithmetic mean, which can be skewed by a handful of highly cited papers.

The whole point of forking is that there's something you don't like about the project you're forking from. As long as that's a technical decision rather than a political one, supporting both old and new versions undermines that technical justification, since it sticks you with all the problems of both versions. Not to mention that it adds the complications of making swapping possible. It's a terrible, terrible idea.

What if a surge takes both of them out?

Or user stupidity erases the vital data? Or malware starts corrupting your files? Or a disaster destroys the whole computer?

RAID is a great solution to hard drive failure, but it doesn't cover all of the other things that might go wrong. For that, you need a proper off-line backup that can protect you against user or OS problems, ideally one that's located far enough away to recover your data in the event of a disaster. RAID is best in addition to, not as a replacement for, true backups.

I suspect that some of it is that the Supreme Court is setting binding precedents for the whole country, so they want to make the logic and implications of their rulings as clear as possible. If any other court in the country writes a bizarre, illogical, or incomprehensible decision, it will get straightened out on appeal by the next higher court. If the Supreme Court writes one, the mess winds up back in their laps in the form of contradictory rulings by lower courts that they have to straighten out. That has to provide some motivation to make the rulings as clear as possible.

It might not be that bad. Yes, if suddenly everybody had a perfect lie detector, people's feelings would be hurt when they discovered how people really felt about them. But after a while, everyone would get used to hearing others' honest opinions. We'd all grow thicker skins and have a better idea of what we're really like.

You can specify the versions of Android your app will work with. Developers set upper bounds and then forget to update them

But that doesn't make any sense. If Android is so amazingly backward compatible, why even bother specifying the highest compatible version? If you are going to have one, why let the developer set it- and get it wrong so often because they don't bother to check- rather than having the Play Store check compatibility with newer versions of Android as they come out? It's just a mistake waiting to happen.