So what's available in Poland? How much do they mark it up? CREE is American, which I think is wonderful and amazing, but I don't think the Chinese stuff will be crap for long. And how much does power cost in Poland, per kwH? That's what makes all the difference.

Where the heck are you? Home Depot stocks the Phillips LEDs, and all the parts I use are available mail order from an outfit in Vermont. There are others, these are just the guys I stumbled across when I first set out to build bike lights.

This stuff is not-not-not prototype -- I built my first set of lights in 2008 or earlier: http://dr2chase.wordpress.com/2008/10/19/more-undercabinet-lights/

The biggest problem is that the design point for incandescent bulbs is all different from LEDs, and trying to put LEDs into a compatible package in a compatible fixture is a PITA. When you can avoid that (under cabinets, for example) results are far better.

Power cycling does NOT kill LEDs dead. Where do you get this information? LEDs are installed on bicycles running on one phase from a bicycle hub generator; at low speeds, it is flicker-flicker-flicker. Chopping LEDs at a kHz is a recommended way of modulating their power. LEDs are used for brake lights (and now, headlights) in modern cars; those are cycled frequently.

The Phillips bulbs are notably NOT harsh; they're a low-color temperature light. I personally like a hotter (bluer) light, but that's not available yet in a good screw-in bulb (Home Depot has some other high-powered brand X that does a nice impersonation of a welding arc; THAT is harsh. Don't buy that one.)

The neighbor post is an idiot. Modern high power white LEDs deliver a much more even spectrum than your standard fluorescent bulb. It's not black-body, but the LED I can buy at Home Depot is far better than any CFL or fluorescent tube I have ever bought anywhere (someone elsewhere asserts that very good fluorescents can be had, and I'm willing to believe it). If it's my own work -- mixed color temperature mounted under cabinets over a counter, I beat that handily. For example: http://dr2chase.wordpress.com/2011/02/20/undercabinet-lights-basement-kitchen/ Yes, there is a bit of a dropout at 480nm -- I know that was immediately obvious to you -- but if I cared, I would fill in with blue+cyan.

A crazy subset of US people are hoarding those light bulbs. Home Depot had the Phillips LED screw-in replacements on sale for $13 yesterday. No mercury, decent color temperature, expect them to last at least 25000 hours if you don't use them in a closed-up fixture, and probably longer.

And you can always find some fool, ignorant of heat pumps and the inefficiencies of generating electricity, defending crappy old incandescent bulbs as a good source of heat.

I know it's done that way now, I was wondering if there are any advantage from the POV of delivering a diffuse light, moderating the blue peak, choice of phosphors, or improved cooling. I'd be surprised if being on the substrate didn't bring some constraints with it. This would be useless for the handy-flashlight market, but that's a tiny market compared to lighting in general, and once LEDs take over we might change how things are done.

I think you're being misled by the height of the spike at the blue end, versus the (lack of) width. And if you had a powerful need to fill that gap around 480nm, they make LEDs around that color, called "blue" (470nm center) and "cyan" (505nm), so you could mix 5 (cool, neutral, warm, blue, cyan) and get pretty good coverage. In particular, Luxeon Rebel, Blue, bin codes 4 and 5 -- 475-480 and 480-485. This is really, truly, not an insurmountable problem, if you really want quality.

Fluorescent bulbs, even high-quality ones (see below) have the problem that their basic light is not just "around" particular frequencies, it *IS* a small set of particular frequencies. You can see this with a diffraction grating: http://dr2chase.wordpress.com/2008/05/08/spectrum-led-vs-fluorescent/ Interestingly, you can see the 480nm dropout in the LED spectrum (narrower part of the smear) and you can get a feel for what color that is -- truly, blue-cyan.

I am a little curious whether it would be possible to make a two-part "white" LED for diffuse light; have a separate glass with phosphors on it, and illuminate that with royal blue.

Thank you for that, and I think I need to retract my remark somewhat. I may be buying some of those, because screw-in LED is picky about being in some ceiling fixtures (because LEDs really don't like to be hot). It's damn annoying that the market for lights is so heavily skewed towards "cheap".

In reply to a previous Slashdot article on LEDs, this minor effort: http://dr2chase.wordpress.com/2011/02/26/led-color-rendering/
The summary is, if you take decent LEDs (CREE or Luxeon) and mix the color temperatures (warm/neutral/cool) it's not bad. Your eyes adapt; the camera is much less forgiving.

And decent LED kicks the crap out of fluorescent.

You could go look at the data sheet, and see that these LEDs don't emit any wavelengths shorter than about 410nm, and the primary color appears to be "royal blue" (445nm).

No need to be "concerned".

You have to be a little careful about the skew from urban heat islands (yes, I know this is a denialist meme, but it is also a real effect, and you would not hear "white roofs" and "white parking lots and streets" proposed as a mitigation if it weren't) but we're getting many more high temperature records broken than low temperature records. You would expect these to occur in about equal number on a yearly basis (strictly speaking, on an 11-yearly basis because of sunspot issues, and there's also weighting that comes from El Nino/La Nina events, but the guys who report all this data are experts and take this into account).

You've got to be careful about interpreting a lot of this data casually; flooding has been observed to get worse in urban areas, and (according to a presentation I attended recently) this has almost everything to do with pavement and not much to do with climate change. A few years ago we had a 50-year rainfall paired with a 200-year flood.

Read about universal hash functions (the writeup on wikipedia is not that bad). They're not a hack.

You don't necessarily use a small space, either -- a 64-bit hash is not normally regarded as a small space, thought it is often smaller than the bit size of what is hashed into it.

Two problems with trees are that you need to define a comparison (you can often concoct one, but they're not always given to you) and though memory is cheap, *probes* into memory are not. If a hash function can get you there in 1 step with high probability, that's interesting.

True, but good random numbers (good hashes) have interesting and powerful statistical properties.

Carefully chosen file names (a lot of them) can DOS file system performance. Whether this could be escalated to a network vulnerability, hard to say -- if an attacker over the net can figure out a way to induce particular file names on the server, that would be worse.

It's a little sad that people are still forgetting about this failure mode of hash tables and hash functions; either there's got to be a randomizing secret swizzled in, or a better (more nearly cryptographically strong) hash function, or both.

Right, but I was talking about existing nuclear plants. We don't avoid decommissioning costs by shutting them down early because gas is cheap and safety upgrades are expensive. It's possible that decommissioning ten or twenty years from now might be cheaper; one of the claimed advantages of the thorium-based reactors is that they can "burn up" the leftover crap from existing reactors.

Pretty sure a big contributor to reduced emissions is cheap fracked gas replacing coal in power generation; for a given amount of energy, it produces less CO2. This is not the result of some grand strategy, and it has inherent limits (there's still a C in CH4), and we'll run out of gas someday.