It’s possible that the problems won’t be as bad in GNU/Linux if the security model is more robust and holds up better, but then again it’s possible that model will have to be debased in order for more people to adopt it as “user friendly.”

Yeah, ouch. I’m not looking forward to the learning curve myself.

Just to play devil’s advocate for a moment:

people somehow aren’t counting all the spyware and virus infestations when they think about ongoing difficulties

Macs don’t face the same problems either, for the same reason: malware authors target Windows because of its marketshare. If Mac or GNU/Linux or anything else ever achieves, say, even 25% marketshare, it will start to draw the attention of undesirables with coding skills.

And I’m sure you’re familiar with the idea that Windows isn’t so easy to set up and keep running either, it’s just that it comes preinstalled on most machines, and people somehow aren’t counting all the spyware and virus infestations when they think about ongoing difficulties.

Where I think the block lies is the perception that GNU/Linux is a dog to set up and keep running. That’s still my feeling, though I acknowledge that it’s way out of date. I really should just get an old machine and try to get GNU/Linux up and running on it. I even have a Red Hat distro around here somewhere, and where I work there is a room packed full of (not very) old computers. I’ve been saying for years that there’s a fortune to be made by selling computers packed with free software to scientists…

@Nemo: Interesting points, and right on about the difficulties of searching when you don’t know the answers, and the problem with people who have the answers often not understanding why it’s difficult. You can eventually get there, but it may seem like too much trouble for a lot of people we’d like to convert. (I need to write up my experience trying to figure out how to use remote drives in GNU/Linux.)

@Taran: I think there are a lot of very friendly and helpful people out there. Maybe I tend to see the scorners because I enjoy a bit of controversy and back and forth. (Not that I’m the one raising it, but I’m interested in both sides of the divide.)

@Bill: Absolutely. I also think free software would be very appealing to scientists that need a lot of crunching done but don’t want to pay for licenses to run Windows on a large number of boxes. I think we have free software equivalents covered pretty well for the apps you mention, but aren’t there a lot of specialty/proprietary scientific applications that we might have trouble replacing?

I think the free software distributions have come a long way in the past few years and that new users should now have little difficulty in starting with GNU/Linux, except for missing out on a few proprietary things like YouTube flash videos, or crummy web pages that require IE. I think there are a lot more problems for people like me that have some history of Windows abuse and a shopping cart full of junk we want to bring with us.

If there were a GNU/Linux bundle that would replace (and be compatible with) Powerpoint, Word, Excel, IE/Safari, Acrobat and Photoshop on Windows or Mac, AND could be reliably installed on a majority of computers that would otherwise be scrapped (that is, models from a few years ago), I think the potential “market” in research would be huge. It appeals to researchers not only because it’s not proprietary, but (in fact mainly) because it’s so much cheaper.

Convince the world’s scientists that it’s safe to thumb their noses at Microsoft — that GNU/Linux will not be more problematic than Windows already is — and you’re on your way to a big expansion in marketshare.

Maybe they wait and ambush newbies? :-)

Search engines work very well when you search for the answer, and very poorly when you search for the question.

Result, of course, being people search everywhere for their question, don’t find anything, post to usenet, and get instantly flamed by three or four people saying JFGI (Just … google it) and posting various queries that never occurred to the searcher to try. Because for someone who already knows the answer, finding it with a search engine was easy, and somehow they think this means it would have been just as easy for someone who only knew the question.

Oops.

Ultimately, there are two sources of information on the internet: googling it and asking someone. When the first fails the second happens. We need more tolerance from question-answerers, but I think we also need better search, or at least better search-optimized pages. Pages with the answers to important questions should be made to score highly on queries based on the questions, even where those queries are phrased entirely in the problem domain without words from the solution domain. Otherwise they won’t be found by those who need them most.

This can be solved by making pages that are the answers to important questions score highly on problem-domain terms. Fortunately, with modern search engine ranking algorithms, that doesn’t require the page author’s cooperation. Anyone who finds it useful to answer a given question can link to it using the question’s language in their link text and enough people doing so will boost the answer page’s ranking for such queries.

Smarter search engines would, of course, help too.

And so would better documentation bundled with something.

Five years ago, this experience was fairly typical:
* Download some free software.
* Install it.
* Run it.
* It works, but there’s a lot of documentation it’s supposed to have, which isn’t showing up when you go to browse it.
* Look about the readme.txt file that is available.
* Apparently the documentation needs to be “built” to be usable.
* Perform build procedure, thankfully not documented in that documentation that still needs to be built but rather in the readme file.
* Procedure fails, saying the sources are missing.
* Reinstall. Same deal.
* Check the .rpm file manifest. The sources ARE missing.
* Where to complain/get the missing files?
* Seems there’s a mailing list and no other alternative.
* Sigh, another login/password pair to remember.
* Create one. Go to post. Oh, what’s this? Suggests searching the archives for your question first.
* Check archives. Great. They’re full of email addresses, all of them in clear text. And the answer isn’t already posted.
* Sigh, write email, and resign self to a noticeable increase in daily spam.
* Get flamed, and caught in a circular finger-pointing exercise where the developers blame the packagers for omitting stuff, the packagers blame the developers, and neither side offers you any clue as to how you can obtain the missing files. The most helpful suggestion being to *write* the missing documentation yourself — which is ludicrous, since you’d need to get the documentation first to know what to write, and if you knew it all backwards and forwards anyway why would you even be asking???

Part of the problem was to lose sight of the fact that there was an end-user with a problem who wants to download a file, push a button, and have their problem go away, not actually do a whole bunch of extra work or even create manually the stuff that was omitted from the distribution. Especially given the implicit assumption that the user would even have the skills to do the latter.

If I got a flat-packed furniture kit from IKEA that was missing a part, IKEA would exchange it or furnish the missing part, not argue with me between whether the factory or the shipping company was to blame, and certainly not suggest that if I couldn’t find the part I should actually make it myself from raw materials! And IKEA behavior is typical of when open source does something *well*, but is hardly the apotheosis of user-friendliness itself…