Most people bin them, just like most people bin batteries.

But (at least in the UK) most supermarkets have battery recycling bins - it wouldn't be hard to add a CFL recycling bin to encourage recycling once they die.

If it dies after a week, then return it on your next trip to the shop, which will undoubtedly be within the warranty period of the bulb. It's the cheap electronics they put in mass produced CFLs, that's why some die early. They seem sensitive to less than ideal electricity supplies (and temperature).

The worry is that LED bulbs and CFLs share very compact electronics, and that is what dies most of the time. Hopefully the LED bulbs currently have higher quality electronics being a premium product at the moment.

Overall, the electricity savings make up for the additional cost many times over, even if the odd bulb dies too quickly.

Nope. How can an inert thing be "evil"?
Nope. They're pretty cheap these days. Not as cheap up-front as an incandescent, but the savings in power more than make up for it.
Nope. If you're talking about the trace amount of mercury, you need to consider the mercury emitted by the power station burning coal to power your 100W dinosaur of a light.
Nope. Many people have 10, 20 year old CFLs still running.

Stop basing your opinions on cheap-ass CFLs sold by cost-cutting, quality-cutting retailers.

The problem with CFLs and LEDs is that they incorporate AC/DC electronics in every bulb. It is this that usually fails, rather than the bulb. Obviously, with electronics, it usually dies quick, if it's going to die, and you can take it back to the store for a replacement under warranty. The solution in the long term is that houses should have DC lighting circuits, and one, high quality, AC/DC convertor for all the sockets. Fine for new builds, but nobody will want to rewire, so the electronics in every bulb is here, probably for a long long time.

And yes, occasionally an incandescent can last a long time. And for all that time, it is burning 60W or 100W, rather than ~20W. And for this reason, the people who benefit most from up-front expensive bulbs are the poorest, as the ongoing running costs save far more than the up-front cost.

(In 100 years we will probably have people running 100 year old LED bulbs too).

LED light quality is surprisingly good as well. IMO more agreeable than CFL.

You can post all the facts you want, but some people will still be whining about their precious dangerous hot wasteful incandescents.

Maybe you should try an infographic version of your post.

Oh, and don't forget the lifespan of the LED bulbs. On a per-year basis, and an average of 15 years (or 25 years) lifespan, you're talking about 50 cents a year for the bulb itself - probably lower than the cost for incandescents given how frequently they fail.

You won't be able to convince some people that this is a problem.

Yeah, I know (having used Linux for 15 years) to untar the tarball, how to untar a tarball, where to put it, how to symlink it, etc.

What we need is a downloaded script that add a PPA and apt-get installs the software using that PPA, and thus integrates that software with the standard application management software. Not that this wouldn't be a shocking way to introduce a million horrible bits of malware, etc, onto a system. Maybe third-party PPA systems could be authorised by a distro as being safe somehow, using certificates or something.

TFA wasn't working when I tried to read it.

They're lucky that they found some cheap monitors, but that's not going to work for everyone.

It's a shame that they couldn't compare their experience of 4K monitors with the alternative of dual 1080p monitors (which would run around $300 today). But yes, they do get the screen area of four 1080p monitors with this solution.

These monitors are probably a bit more than $500 each at the moment.

But if your developers are $100k each a year, $1k or even $2k for a three to five year lifespan is not a problem, especially given the productivity increases claimed.

But if their previous monitors were fairly new too, then they are splurging money a bit.

It's not really that much worse than having two monitors side-by-side (e.g., I have two 22" monitors), except it's a single contiguous space ... and twice as high.

I'd probably want to put the monitor further back on the desk though otherwise it would just be too imposing.

I think you read the game size in bits, not bytes.

20 Gbit: It's under six hours: http://www.wolframalpha.com/input/?i=20+billion+bits+%2F+1+million+bits%2Fsecond+in+hours (But that assumes you don't get slower speeds at peak times.)

20 GB: http://www.wolframalpha.com/input/?i=20+billion+bytes+%2F+1+million+bits%2Fsecond+in+hours (45 hours)

A Tottenham Hotspur fan will use the term "Yid" as a "reclaim the word in an empowering manner" way. I.e., they can use it, nobody else can. The FA should recognise this in their guidelines. Yes, a lot of Tottenham fans aren't Jewish, and I'm sure some of them don't even know what the term really means. But they are not using it in a negative, insulting or defamatory manner. I appreciate that in a different country, the term could be seen as a far worse word.

The fans arrested in this story were opposition fans, mentioning Auschwitz, etc. Straightforward racial hatred.

Yeah, and Amstrad released the 8-bit PcW16 in 1995 for whatever crack-addled reason they thought it was a good idea. 16MHz Z80 sure, but still...

Yeah, the C128's video chip (MOS 8563) had it's own 16KB (up to 64KB) of memory so it could operate completely on its own without affecting the rest of the system, to generate its own display. Very clever. The downside being, of course, that the video memory wasn't directly accessible by the CPU, all operations had to go through the video chip.

This wasn't particular unique of course, MSX video chips operated the same way, and the 8563 did have a primitive memcpy hardware to aid in memory manipulation.

Odd that there were so many Z80 systems that shared memory with bitmapped graphics that never had graphics corruption issues that you mention. I guess you are talking about the late 70s, because you're talking about the 8080, static memory, etc. The C128 came out in 1985. The C128's MOS8563 was a character based CGA-alike chip that you are dissing in your post (and like the MC6845 that was used for character based CGA displays, it could be hacked into being a bitmap displaying creature).

The speed of a 4MHz Z80 matched well in reality with a 1MHz 6502. Yes, the 6502 had better IPC, but the Z80 ran far faster to compensate. There were 2MHz 6502s too, as used in the C128 and BBC Micro, these should have allowed these computers to be noticably faster than the Z80 systems, but they weren't.

Actually, it's really odd reading your post, you are mixing up mid/late 70s computers with early/mid 80s computers, at a time when every year brought major progress. I don't see the point of the comparison.

Well it took until the A600 for C= to make a sub $200 Amiga to replace the C64. The $1700 A1000 was not a serious option at the time for people that needed a computer with a software ecosystem.

In 1985, a $300 C128, or a $500 C128D, made a lot of sense for a home computer - a C64 for the kids, a CP/M machine for the parents.

But like all "multipurpose" home computers, it ended up being a C64 for the kids only!

Well, it didn't take off in the way the C64 did, but it still sold 5 million devices, generating $1.5b in revenue for Commodore to waste on bad decisions. Bill Herd says they expected to sell 2 million devices, so in these terms it was a success.

And consider that this was available for $300 too. Yes, a big bundle of chips on a rammed motherboard, very much unlike the far cleaner designs coming out at the time.

It is really odd that not much C128-only gaming software was made - I can only assume that for games this was because the C64 version already worked on the hardware, so why spend extra effort making a C128 variant that actually made use of the faster CPU, extra RAM, and so on? And it was probably hard to compete with the CP/M market for dedicate C128 80-column business software.

Reading the specs of the 80-column chip, the MOS8563, it appears to be very much like the MC6845, but with attributes and a simple blitter.