There is a traditional solution to this, in theory. The theory goes that a newspaper will meticulously edit and fact check, because at some point, they are bound to write about something I happen to know a lot about, and then if they are wrong, I will no longer trust them on other things. Unfortunately, the theory works a little too well: virtually all newspapers, if you read a few issues cover to cover, contain enough errors about things that you happen to know a lot about that you start mistrusting them. But there are a few publications for which traditional solution seems to work. I have rarely found any mistakes in The Economist, for instance, when they write about things I happen to know a lot about, which leads me to trust them more than most other news sources when it comes to things I don't know.
Maybe the reason that IT Conversations still sticks everything into a single RSS feed is that they are secretly hoping for somebody else to do the fancy classifying and tagging for them...
As I understand it, traditional search engines already use the feedback of which links people click on (I think they used to mangle the links in order to be able to track this, but it seems that at least Google has now switched to some JavaScript trick.) If the goal is just to improve search quality (and this may not be the case, I'm not sure), then whether you'd reach that goal will hinge on whether you can get more accurate information by letting people vote explicitly, and after they have already seen the page, rather than implicitly, and after they have only seen a short summary.
But it will be interesting to see if some of the business practices (very low initial budgets, early releases, &c) common in web startups might translate into other fields. At least one YC startup (sproutsys) appears to be doing pretty serious systems software, which is still software, but quite different from being a "web startup".
I imagine that hiding the screws is probably great for attracting some segments of customers, but it repels others (namely: me). I don't know if there are many people like me, but one of the reasons why I stopped buying Apple computers and switched to IBM/Lenovo is because the Apple stuff, despite being quite reliable, would inevitably break down at some point, at which point it was nearly impossible to take it apart without first reading and rereading the secret licensed-Mac-technicians-only how-to-take-it-apart manual.
Apple tends to optimize for ease of learning by beginners, not ease of use, let alone tinkering, by experienced users. I don't mean to be patronizing, I'm just observing. In my opinion, truly great user interface design provides a smooth learning curve that's easy to get onto for Aunt Tillie, but also continues to be smooth for J. Random Hacker.
Products with that kind of learning curve are great from an abstract engineering-aesthetic point of view; but assuming that your own sense of aesthetics translates into profit is, of course, wishful thinking. Still, I'm looking forward to a mythic future in which consumer electronics will again be more like the Apple II, and less like what Apple makes today. Examples going in that direction would include the OLPC laptop (http://laptop.org) and the Chumby (http://chumby.com) .
You can also consider setting up a single proprietorship or an unlimited partnership for the time being. That's a lot cheaper in Holland, and I imagine also in most European countries. The only big problem is that you cannot borrow money without being personally liable, and you cannot sell shares, but if you just need a "business" for the time being for tax purposes and to order stuff from places that will only sell to businesses (more common in Europe than in the US), it will do fine. In fact, I once set up a single proprietorship for the sole purpose of being able to buy food for cheap at a restaurant supply store, and it cost me about half of the Oregon LLC referred to earlier, and was done in a half an hour at the Chamber of Commerce, without lawyers involved.
I submitted this post to news.yc because, even though it is phrased as a feature request to an established company, there is a lesson to be drawn for startups. If Google had made it easier for me to enter metadata in bulk, they would have gotten some valuable content and attracted more users.
Reddit needs to figure out how to create more subreddits without having to overcome critical mass every single time. I'm sure I'm kicking in an open door here, and I believe I read that they were working on it already, but I haven't seen a real solution so far. The recommendations system theoretically makes it feasible for all articles to go on one big reddit even with a large and diverse user base, but in practice, they're still suffering from a situation where new subreddits are hard to get off the ground while the main page has to be shared by different subcommunities with different interests.
If you are anywhere in the EU you can probably start a British limited company, even if helpful government officials say you should register with the local overpriced companies registry. In The Netherlands there are several law firms that specialize in this, and they charge about EUR 500. I imagine lawyers in other EU countries have figured out the same trick by now. One thing to keep in mind is that you may expose your company to being sued in the UK, but you probably should be avoiding getting sued anyway.
You say that type mismatch errors are orthogonal to testing. This may be the case in languages like Java, which has a strong but not very expressive type system. In strongly typed functional programming languages like ML and Haskell, though, you can, with a bit of good design, capture a lot of information about what you want your program to do in the type system, to be checked automatically. Take, for example, the comment on this page by somebody who says that the only testing s/he does is XHTML validation. With strongly typed functional programming, it is possible, and not really even all that hard, to write an embedded combinator library for XHTML such that programs that generate improper XHTML don't type check in the first place.
OCaml is an object oriented language, yet people who do real work in OCaml (and yes, there are such people, and their work is very real) appear to rarely use the object oriented features at all. The reason seems to be that the combination of a fancy module system with the standard trappings of higher-order, typed functional programming (algebraic data types, pattern matching, and the like) solves most of the same problems as OO in ways that OCaml users find easier to use.
There is, indeed, no such thing as inherent value, but people nonetheless tend to think in terms of inherent value. This is important to realize when you decide on marketing and pricing schemes. For instance, restaurants are more likely to offer a discount on quiet days than charge a surcharge on busy days, and they make those sorts of decisions for a reason (restaurants can be a very competitive business, after all.)
David Friedman wrote a neat paper about the psychology of inherent prices and some related topics:
Good old Bastiat... That's definitely fairly remote from "mainstream" economics these days, although there is a lot of sense to the proposition that the works of the classical economists are more practically useful than the abstract modeling you'd learn as an economics major. If you nonetheless want, for whatever reason, to learn something akin to the modern standard microeconomics curriculum, albeit with a good deal more game theory thrown in, I can recommend the free (Creative Commons!) textbook by Preston McAfee at http://www.introecon.com
Computer Science is an unusually accessible field of engineering for self-learners. Purveyors of fine computer systems and programming languages market them by writing oodles of documentation, often free. There is an amazing amount to be learned just from reading the GNU manuals. All the machinery you need in order to play around with most of the latest new technology is available at Fry's.
But I should probably warn you against the idea that you can become a good programmer simply by learning to use some of the tools. That's like becoming a chess player by studying the rules of the game, and the mechanics of chess clocks. If you just want to make sure you can understand your more technical business partner when s/he talks shop, or even do some simple programming, surely, you can get a long way simply by knowing the syntax of a programming language or two and some basic UNIX. I've seen some self-taught programmers do amazing things. But for most jobs that are actually technically innovative, you're going to have to sit down and learn some stuff about algorithms, data structures, concurrency, and so on. (This is what I shall snobistically call "real CS".)
Fortunately, even "real CS" can, with lots of motivation, be learned outside of formal academia. I would recommend you start with Structure and Interpretation of Computer Programs (the book and three different series of free video lectures are on the net for free). In fact, even if your goal is just to understand the techies, SICP is probably still about as good as it gets, and the old Abelson and Sussman video lectures are quite amusing. Other great resources are ArsDigita University and OpenCourseWare.
I once talked to a TicketMaster recruiter at a job fair. When he realized I was getting a little bored, he told me that besides the positions they were advertising, they also needed somebody who could help them speed up their main database system, which, if I recall correctly, was written in MACRO-32 and ran on a VAX until they moved it to Linux boxen running a proprietary emulator. If you search around the web for "ticketmaster vax" you can maybe find out how far they have gotten with their project to scrap the VAXen.
The list of browsers you can test on includes MSIE 5.0 and 5.5. Do people target these browsers anymore? How old does a browser have to be before you say: I'm satisfied if the layout is a little off, as long as it's legible?
You're right about that, but at the same time, this is also exactly the problem. If you aren't going to be able to take up the Luxembourg market without paying particular attention to the cultural and linguistic peculiarities of Luxembourg (hint: 3 languages), then the conclusion for many companies is going to be that it's not worth worrying about Luxembourg at all. Then again, if nobody worries about Luxembourg, that opens up a decent opportunity for somebody to just copy a bunch of existing ideas for the Luxembourg market. And that's why people are looking for ways to overcome the problem of cultural fragmentation and make it profitable, even for a small company, to target the European market.
Many European countries have minimum capitalization requirements for limited companies, as well as high transaction costs (you've basically got to pay the notaries' cartel their monopoly rent) and background checks. Holland got rid of an EUR 18k (!) minimum cap requirement a few years ago, after a court decision made clear that it was perfectly legal to incorporate in England instead. Even now, the transaction costs associated with starting a Dutch BV are easily twice what you pay for an English Ltd., and reporting requirements are more complicated too. The Chamber of Commerce, a QUANGO that is supposed to advise new companies and which doubles as the registry of companies, likes to claim that a Ltd tends to have a harder time raising capital than a BV, presumably because it sounds weird and foreign; but I suspect that the fact that you pay a low fee to Companies House in London instead of a high fee to your local Chamber of Commerce is also part of the reason why they discourage jurisdiction shopping.
Metaweb (freebase.com) uses (or rather, is) a database, but not a relational one. Those comments that implicitly identify "database" with "RDBMS" may become outdated sooner than their authors expect.