I did research on this stuff back in the 1990s. Made the films, did the vacuum chambers, had the world record for emission efficiency for a while. While it may have some niche applications, the basic problem is that it is *not* a low-voltage technology. Modern chips operate on 1.5 V or so; Diamond devices will be more like 5V. So ultra-low power? Nope. They say that the devices are more efficient because the electrons don't bump their way through the silicon crystal lattice. While that's true enough, it doesn't actually make a big difference. Why? Because the electrons very much will dump all their energy when they leave the vacuum and hit the anode.

Ultra-high speed? Again, while vacuum is nice in that it doesn't slow down the electrons, that turns out not to be a big effect. The most important factor in speed is the size of the device, and there is certainly no reason to believe that these vacuum tubes will be smaller than transistors, if built with the same lithography tools. I may be wrong, but I have good reasons to believe that they will be harder to make small.

High temperature? Radiation resistance? Maybe, but that turns out to be a complex question. These devices aren't just diamond and vacuum. They involve insulating layers, too, and those insulators may be affected my high temperatures or radiation. Essentially, a device is as robust as its weakest link, so until you can make the entire device out of truly robust materials, you won't gain too much.

So, it's nice work. I know how hard it is to do this stuff. And, it might be useful eventually. But it won't revolutionize technology any time soon. And, those guys ought to realize that, if they would let themselves. Research lives off publicity these days, because it is being forced to become more and more of a competition between groups. The trouble is, when competition enters and your salary depends on the claims you can make, truth tends to be (shall we say) over-inflated.

That darn free market ideology messes up science. I like it as much as anything for people who make spoons or telephones. But science isn't making spoons. If you get a bad spoon, you'll know it, but if you read an exaggerated research paper, how can you tell, other than by doing the research again? And, that's just not efficient: doing it wrong and then doing it again isn't nearly as good as doing it right the first time.

Oh well. Enough ranting.

Nope. Sorry. You're wrong. Electromagnetic waves add very nicely, so that your signal remains there, even if many other signals are simultaneously being transmitted.

The overall idea is fine, in principle. As other people have said, it is 802.11n beam-forming on steroids. If you had 1000 transmitters, and if you could know the exact time delay and attenuation from each of those transmitters to your cell phone, then (indeed) you could make them all add together precisely where your cell phone is. Elsewhere, they would (on average, mostly) cancel out. As advertised, you would (indeed) get a substantial improvement.

But not a *huge* improvement. With N=1000 transmitters, the signal at your cell phone would be (e.g.) 1000. Elsewhere, the signal would be (in the same units) typically sqrt(N) or about 30. That value of 30 is the noise level for eveyone else, and their transmissions give a similar noise level to you. So, 1000 transmitters gives you a signal-to-noise ratio of 30 times better than one transmitter.[*]

[* ...actially, if you're really gonzo, you can adjust all the transmitters to make their signals cancel out exactly, at everyone else's cell phones, so long as you have more transmitters than cell phones. In principle. But I don't think anyone is seriously proposing that...]

Now, you plug that into Shannon's equation, where the data rate is B*log((S+N)/S), B is the bandwidth, and S is the signal and N is the noise. If you start with S=N=1 which is not too far off for a normal system, and change it to S=1000, N=30, you find that the data rate increases by a factor of 5. Which is good and impressive.

But the problem is that it is very hard to know the attenuation and time delay precisely.

So, anyway, it's not a bad idea, but it's hardly new. It's also not easy to implement, and even if you implement it, you won't get factors of 1000 improvement.

Yeah, except that lowering taxes doesn't seem to grow the economy. Certainly not enough to make up for the lowered rate. Too bad. It's all sensible except for that point.

Anyway, the top 20% pays 86% of the taxes, maybe, but don't they own an equally large (or larger) share of the wealth and income?

When I was a kid, in the 1960s, tax rates were way over 50%. Tax rates in England peaked at 90%. While 90% marginal tax rates will certainly hurt your economy, it's not at all clear than 30%, 40%, 50% rates will do much damage. If we can manage to spend some of that on useful infrastructure things, it may even be good in the long run. Just think! We could have a well-educated, healthy workforce, streets and networks that work, et cetera.

Well, find out for us instead of just talking. Write the code both ways and show it.

Anyone with *any* experience of software knows that you cannot tell if something is "easy" or not until you've done it (or done something that's very similar).

That's all I can say.

Could be a boring life. I suppose you get tired of them rapidly because you don't really know them all that well.

Sounds like you're living your life in fear of losing half your property.... Weird. Property is just stuff. Plastic, mostly. Soon to be junk, mostly. Stuff that requires you to buy closets. Stuff just gets in the way, mostly.

One of my most enlightening experiences was when I moved to England and found out how small the typical English house was, compared to my American house. I ended up throwing out/selling half of everything I owned. And, guess what? I didn't actually need any of it. (Of course, I would have regretted losing some of the *other* half...)

That's a very self-centered view. Not evil, but just thinking about yourself.

Some of us like the knowledge that there is someone permanent around who is willing to help. Someone to talk to. Someone who gives a shit. You can't buy that last two. You can buy a helper, you can buy sex, but you can only buy people who pretend to care about you or about what you say. If they are in it for the money, they don't care about you. Rather, they only care about your money.

And, some of us are willing to actually pay attention to someone and to give a shit about them, in exchange. You can call it love, you can call it partnership, you can call it whatever you want.

Absolutely. By blocking anything potentially dangerous, you end up with a safe organisation that isn't able to function well.
Obviously, the I.T. guys see their own pain. But, the pain that excess security causes is widely distributed across space and time, and no one counts it all.

So, in this case, yeah, a virus is bad news. But, the question is, is a virus more lost productivity than 1000 people who are unable to send zip files?

Yeah, right. Scientific software is not at all like a word processor. In a word processor, you can tell immediately if it is behaving wrong: you know what it is supposed to do. But in some scientific computation, it's just the reverse. It tells you that the answer is 3.184, and it is not immediately obvious whether the answer is right or wrong. That's the difference.

"To weed out the crap" as you put it, you need to understand the computation in detail, design some test cases that are relevant to your own problem, test it, and think about the approximations that are being made. This can take weeks, months, or even years.

And, then after you've decided if it works well enough for *your* research problem, the next guy is probably going to run it on a slightly different problem. Some of those questions will raise their ugly heads again. Are those approximations still good enough? Do I need different test cases?

Basically, the difference is that with commercial software, you write it once and people run it a billion times. With research software, you write it once, and run it once. Maybe, you'll use it again, once, or maybe not. Maybe three other guys (girls) will use it too. But that's about it. Once the program works, your research question is answered, and that might be the end of it.

Oh, so young to be so cynical. The interesting thing about real science is that *sometimes* the answers are unexpected/unwanted. The other interesting thing about real science is that you actually find some scientists who care more about the real answers than the answers they were "supposed" to get. The third interesting thing is that there are nonscientists and funding agencies who actually care what the real answer is.

So, sometimes, cynicism fails. Don't be corrosive.

Exactly. The phrases used are fairly long, for instance: "Laugh, dance, and sing if fortune smiles upon you." In the TIMIT corpus, there are 122 of them. In the English language, there are hmm, lots of sentences of that length. There are about 1000 different syllables in English, and I count 11 syllables in that sentence. Thus, there are some fraction of 10^33 sentences of that length.

So, if you tried this on English, one of two things would happen. If you used that recognizer without any modification, then it would sit there silently until you said one of the sentences in TIMIT, like "She had your dark suit in greasy wash water all year." And, it would be a *long* wait.

Or, you could change the recognizer so it could recognize more than 122 possible sentences. In that case, the error rate would go way up.

When considered from the point of learning something, there's a certain grey zone. Looking up an answer can be a way to learn, as long as the idea passes through your brain en route. But it's certainly cheating if it goes from the browser to the word processor and you don't understand what you are copying.

Ya gotta cache in the brain.

Yes, it's true you can become an expert in anything these days. I've done it, going from one research field to another, and I'm expert enough to get paid to do the research. But you still have to know a lot of stuff, because it takes 300 milliseconds to pull a fact up from the ol' grey matter and it takes anywhere from 30 seconds to 30 hours to find it on the internet. If you need the internet to answer basic questions about your area of expertise, you aren't an expert. Or, at least your a verrrrrrrrry sloooooow expert.

Geez, guys. Mental age should be at least 16, mental altitude should be at least a foot above the gutter.