http://www.openbsd.org/cgi-bin/man.cgi?query=strncpy

http://www.openbsd.org/cgi-bin/man.cgi?query=malloc

They don't just explain what strncpy & malloc do. They also explain how to use these functions correctly and securely. With examples.

Compare that to glibc's manuals (pretty typical of what you can find under Linux):

http://linux.die.net/man/3/malloc

http://linux.die.net/man/3/strncpy

Clearly OpenBSD folks care.

Hey now, careful with the blanket statements. Some of us hackers actually do care deeply about usability and aesthetics, but "clean" doesn't always mean "better" - despite many UX people's unfounded insistence.

I work for a company that shall remain unnamed, but we do a ridiculous amount of testing for every little UI change. Conversions, sales, and a whole ridiculous slew of variables are tracked in extensive A/B testing for every UI design.

Our app doesn't look like much, but it's also the proven design - and we have the numbers to back it up.

I'm all for usability and aesthetics, but when you're arguing that your clean-looking design, which follows all the usability theories in the book, is better than one with a long track record of successful metrics, you better have some numbers of your own.

In the end Zappos' business is not clean, pretty UI. It's generating sales. If clean, responsive, usable UI leads to that (and it most certainly does - within reason), then they ought to do that. If it doesn't, then it'd be a dogmatic waste of resources that can be very damaging to the company.

First, Cassandra uses a disk layout similar to the one described in the Bigtable paper (http://labs.google.com/papers/bigtable.html sections 5.3 and 5.4); in particular it does no random writes. Relational databases like mysql pretty much all use btree-based storage which was great 20 years ago but is terrible today when seeks are your bottleneck.

I was talking to some people today who are struggling to get mysql to do ~100 insert/update operations per second. Cassandra will easily give you 10x that -- _per node_.

The second benefit is that Cassandra gives you real, scalable partitioning, invisible to the app, for when you do need to add nodes. When you have more than a handful of machines, not having to babysit replication + partitioning is a huge, huge deal.

Astronomers used to use huge glass plates to capture telescope images, but when CCDs became available there was a revolution in data quality. It's true that astronomers probably don't care about the same things photographers do wrt capture medium, but in astronomy, CCDs wiped film off the map.

The final killer for film: You can't back it up. Your slide may last forever, but if you lose it, or your house burns, game over.

Solving a problem a second time is much more profitable. A big part of business is focusing on your strengths, rather than your weaknesses. Find out what you do best and market the hell out of it. No product is perfect.

The point of the article is that the only losers are the ones who try to do anything tangible that requires a collection of patented ideas to function.

Any recording medium has limitations, and whether it's digital or analog is neither here nor there.

If you rescan your 35mm slides at high resolution in 10 years, you'll get a really nice picture of the individual grains in the emulsion, not a better picture.

It generally takes much less water to deflect visible light and reduce the efficiency of a silicon solar panel than it takes to absorb the UV that is producing power through the melanin.

What I believe is important here is the melanin. If an efficient and stable way is found to manufacture a melanin-based solar panel, it could be used in a wider range than many traditional solar panels. I suppose the ideal would be to produce a traditional solar panel that is transparent to UV, allowing it to be caught in a melanin-based solar panel. In theory you could double the power output of a solar panel and extend its power production into cloudier periods.

And I'm not. I'm saying "Amazon is ugly and it works. We have teams of Web-2.0 guys doing redesigns on every facet of the site day-in-day-out, testing with live customers constantly, and this is still currently the best design - and we are still iterating."

A bit of a mouthful, though.

"I'm sure they could measurably prove that customer satisfaction on the web site and perceived brand value would both increase."

That's just it though. Where I work, we have proven numerically that this is a false assumption. "I'm sure they could prove" is a far cry from "We have proven". This is a problem that is prevalent throughout the UX community I think - a dogmatic worship of several principles without ever sanity-testing your assumptions with large-scale metrics, instead focusing on ephemeral and unreliable things like anecdotal user stories.

"It looks like it works better" and "it works better" are entirely two different beasts. One thing you can say off-hand, the other requires backup.

For what it's worth, I was on your side at one time. I hated "dirty" 90s-looking websites like eBay, Amazon, et al, and I loved the new-age Web-2.0-y stuff.

Then I got this job and got a sneak peek into what the user data actually says. Some things defy common logic - or at the very least, user experience design common logic.