I compared physical to e-book prices on all my purchases in 2008 and 2009 and found that my average book price was a little over $6 less than paper. Typical softcover prices are $4-5 for back-catalog stuff, and $7-8 for current, which compares to $7-13 paper. Hardcover/recent release prices are dramatically better: $9-12 e-book (they haven't all be $9.99 in more than a year), $18-22 in paper. And classics are dirt cheap: As little as $0 from the likes of Project Gutenberg, but nicely typeset versions are $2-3. Good luck finding a classic in paper for less than $7 unless it's used.
This is not universally the case, of course; some of the books were about equally priced, excluding shipping, but of course there is always shipping. (I use Amazon Prime, so the shipping is not easy to calculate, but it's there.)
I am a heavy reader -- 2 to 3 books a week. Over the first two years of Kindle ownership I saved more than $1200 versus paper if I'd purchased paper from Amazon. This is actual savings, not made up, I added them up in a spreadsheet in a fit of pique while arguing about e-book futures with someone. But really the savings were much greater: Many of my book buys are impulse, and that means I used to hit bookstores a lot and pay retail prices, especially for recent releases. (As an aside I lament the fact that e-books are the final nail in the coffin of local booksellers. I hate that, although I love having huge catalogs available all the time.)
Of course the readers ate into that a lot; $400 for the first one, $360 for the Kindle2 (because, what the heck, I saved more than that the first year anyway and my daughter can use the old one), and $200 for a refurb Kindle2 after I drove away with the first one on the trunk of my car (this when the readers were still $360). As of last summer I was really only about break-even, but of course every month I go without buying another reader is like another $50 so I'm well up again at this point (plus my daughter's books are cheaper too).
Now, those are all new book purchases, as is my norm. If you're one of those people who hits used bookstores or libraries the economics completely fall apart, although they are getting better as the reader prices drop.
Going forward the economics should only get better.
Dropping prices for e-ink readers are one reason I think the Kindle et al are pretty safe from the iPad. Most of the book readers I know weren't keen on spending $400 for a reader when the Kindle came out, though by last Christmas, at $260, many more made the jump. I think it's a safe bet that you'll see Kindle2-class readers for under $200 by the end of the year, and probably around $120 by the end of next year. It's going to be very hard for the iPad to compete on price. It's a different class of device, so perhaps it will do well anyway, but it isn't going to be mass-market in the way e-book readers are quickly becoming.
Personally I look forward to the competition in e-readers. The more of them that are out there the more competition from retailers and the stronger the incentive to standardize on one book format. (I bet we don't see DRM disappear entirely, for lots of reasons.)
The problem with current e-ink is the horrible flashing when you change pages. Show the text overlain, then black, then white, then just the new text? Ugh.
I was seriously concerned about this when I bought my first Kindle. You know what? Within 10 minutes you just don't see it anymore, pretty much the same way you blank out the page flip on a paper book. Perhaps this is because it really isn't a "flash".
It'd be ok with me if it went away, but it's not a problem or even a distraction.
Why hasn't anybody tried using the yellow-green-magenta-black of printers for a color e-ink display? It should work exactly the same as producing color on an ink-jet printer.
They have -- in fact, E-Ink has been demonstrating the technology for four years:
I concur, but I sure would appreciate having one that was designed to be attached to the reader. You need to hook it to the case on a Kindle, and I have to put it at odd angles to avoid screen glare. Still, way less eyestrain than peering at the iphone.
I notice that the Nook does have a clip-on light, one of several places I think they are superior to the Kindle, but I haven't had the opportunity to see how well the unit works in practice and I'm not convinced that the LCD panel isn't a temporary hack until e-ink technology supports color. It certainly is an expensive feature in terms of battery life.
And to think I bought mine after determining that it was a bloody good handheld internet access device.
The shiny is just gravy; the important thing is that it works really well.
As an internet device anyway. Kinda so-so as a phone, although still worlds better than Windows Mobile. (I haven't figured out what Windows Mobile is actually good at, but perhaps the reason that it has such lousy market penetration is that no one else can, either.)
Considering the iphone as an e-book, it works reasonably well in a pinch. It's nowhere near as pleasant as a Kindle for long text, and I can't for the life of me figure out why everyone thinks swiping is a good way to turn pages when you're doing it every few seconds, and long reading stints drain the battery, but at least it fits in my pocket.
Color e-ink devices will be out next year. I haven't seen any yet, although e-ink was demonstrating them a year and a half ago and people I know said they looked good. I'm dubious because the current screens have that grey tint, and that would make many pictures muddy, but even a little color would go a long way.
In any case many technologies will be employed in e-books over the next few years, you'll be able to pick and choose. I already use LCD and e-ink....
Oh, I dunno, that hasn't stopped people from buying them for Windows. Macafee halves performance on my WinXP PC at work; we have to turn it off on build directories.
I realize that this is near-mandatory pain on Windows, alas, but there will have to be pretty significant threats before I would consider running such a thing on my phone.
LAN/WAN bridging on the WRT54GS peaked pretty close to 32Mbps in my experiments, using DD-WRT, assuming you have the right version of the hardware (pretty sure that was v3).
An Airport Extreme has about the same limit (that's what I'm using now because Apple Airport firmware updates kept breaking compatibility with the Linksys WiFi). In fact, I didn't find any consumer-grade stuff that did better than low to mid 30Mbps range last time I went looking (admittedly a year or two ago now).
I sure did find worse, though: Many of the Linksys WRT54G units could only pull around 14Mbps regardless of firmware. And for a little while I used a Netgear FVS318 but that was awful -- it peaked at about 8Mbps with NAT enabled! (It was what Staples had when the aforementioned WRT54GS decided to die at an inopportune time.) I do not recommend the Netgear; not only is it stupid slow, but if it sees heavy load you get frequent hangs. The only plus was that it was really easy to set up and was a lot better than not having any network.
Immediately blaming Windows and suggesting a 'nuke it from orbit' option like switching OS's? Rubbish. The problem is in human behavior, not in the operating system.
Changing their behavior could make their use of Windows safer, but speaking from practical experience every time I have switched a problem user to a Mac (which in some cases I did at my own expense) the support time fell to essentially zero after a short adjustment period. Importantly, this has always happened. Changing the OS fixed the problem without fixing the user.
Theoretically the Mac (and Linux) can have similar malware problems to Windows, and there is proof-of-concept malware out there. Practically, though, it just isn't the case. I suspect that the security model -- which is much better than even Vista/Win7 when you get into the details -- has a lot to do with that. The claims that it's all about market share are wishful thinking; it's about ease of entry. It's very, very easy to subvert Windows... and very difficult for Microsoft to fix this because a large amount of software depends on the ability to do things that should really not be allowed (like, say, injecting threads into other programs and using VirtualProtect to make code pages writeable -- with those capabilities you will never make a secure system). As long as it's much easier to subvert Windows there's not much point in going after the Mac.
But it goes beyond malware. Certain Windows features, such as the Registry, are hideously overcomplex and unreliable. They should have been tossed years ago -- and since everyone manages registry settings via APIs this could be done with excellent backwards compatibility (consider how easy it was for Apple to switch the format for Prefs in Snow Leopard). When something messes up the registry, all too common in my experience, there is little choice but to burn it down and rebuild from scratch. (Thank God for Acronis or I'd go insane from the reinstalls.)
If you want a system that doesn't require a lot of admin time I have to say that you can pick pretty much anything other than Windows and do well. It will be easier to set up, easier to back up, easier to fix if something goes wrong, and the software will be considerably less expensive (everyone else packages useful software in-the-box).
Of course, there can be overriding concerns that force the use of Windows... and you take your lumps if that's the case.
Ask any random iPhone user if they browse the web. There is a good reason why iPhone users have 5 times the data usage of any other....
Prior to the iPhone I rarely hit the web with my phone (it had to be an outright emergency). The plan made it stupid expensive and I got grey hairs waiting. But having seen how useful the iTouch was when it was on the net I went for an iPhone (AT&T be damned. No, really, damn AT&T). Having a useful web browser pretty much everywhere has been a godsend.
It's kind of a crappy phone, but it's a remarkably good browser despite the form factor. That should only get better across the industry as the devices and software gets more capable.
I don't think that there are many end users who give a whit about the freeness of the code beyond free-as-in-beer. I'd welcome examples to show that I'm wrong, but my experience is that they care about what it costs and not about code availability.
That free-as-in-beer and forced code contribution work hand in hand to benefit the end user is unquestionable, but I don't know many who make that an overriding (positive) concern for picking one product versus another.
I don't disagree that git has some significant advantages here, although I do think that practically speaking they're not that dramatic. I'm always on the Internet when I'm developing[1], so there is little practical difference between the distributed and client/server models.
(By the way, I hadn't known about this video and I appreciate the pointer; sometime when I have an hour to spare I'll watch the whole thing.)
I think cheap branching is more interesting; Perforce branching is too expensive IMO. (I note that there are commercial products that are good in this area, too, like AccuRev.)
The two places I'd be worried a lot about git are in merge management (how easy is conflict resolution?) and scalability. I know git has trouble with the latter, and I suspect it isn't all that great at the former. I have regularly been impressed with how well Perforce manages automatic merges with minimal conflicts that need to be resolved. In a highly active codebase it's hard to overstate the value of reducing the number of merge conflicts that have to be handled manually.
Aaanyway this was just a long-winded way of saying that I still don't think there is a clear winner.
I don't disagree with any of what you said except that AIX is a long way from dead. It's still quite popular, much to my chagrin.
I was going to argue that Linux' low cost was the key to its early success, but the more I thought about it the more I believed that it wasn't low cost alone that did it -- or even that it was the most free. It wasn't the most free, and still isn't; that crown goes to BSD no matter how you slice it, whose license is the next best thing to public domain.
I know Stallman likes to tout the "freedom" aspect of the code, but really the GPL works more like a traditional patent cross-licensing agreement, where parties share each others' technology because otherwise they wouldn't be able to ship a product at all, rather than being truly unlimited.
The GPL takes this to the extreme, forcing unlimited cross-licensing of any derivative work... not only to a select few who can buy into the club, but anyone who has any desire at all to be involved.
This avoids the software version of the Tragedy of the Commons, which BSD has always seen: Lots of products use(d) BSD, but because everyone's changes were proprietary they were almost always kept in-house. BSD development therefore got little advantage from commercial development and the pace of its improvement has been slow to say the least. (It was further hampered, very significantly IMO, by a bad case of NIH in the core maintainers.)
The fact that commercial entities, who had the motivations and money to do major work, had to share their Linux improvements certainly led to major advances that have created a feedback cycle whose power is obvious at this point. This cycle continues not because the software is free as in freedom, but because it is not -- it comes at the cost of technology cross-licensing.
Write once, run anywhere = very cool. Always backwards compatible = very cool.
Hah, we always harped on Sun for that statement. We said it was "write once, debug everywhere." Not so much for server software, with a few notable exceptions where OS differences easily showed through the Java APIs, but UI development was certainly not "run anywhere." (Not that C++ was better by a long shot, but at least when you did it in C++ there was no expectation from your boss that it might actually work.)
As for "always backward compatible"... you must be joking, or inexperienced. I don't think there was a single significant JDK release from 1995 to 2004, when I was focused almost exclusively on Java, where I didn't have to fix a bunch of stuff. Swing in particular tended to find new and irritating ways to not work with existing code. Again server code wasn't nearly as bad, but I think that's because it tended to touch a lot fewer core libraries. I don't recall anything that was difficult to fix, usually just tweaks here and there, but I assure you I got familiar with System.getProperty("java.version").
I see that one of the follow-on comments claims that this is due to poor coding, but I didn't find that to be the case. I saw behavioral differences in libraries that were sometimes inexplicable, although more often it was that they fixed bugs in the libraries. There were a lot of bugs outside of the core classes, especially in stuff like Swing and JavaMail. I never had to write a lot of version-specific code, but it certainly happened often enough that you couldn't haphazardly take an upgrade of the runtime system. You could pretty much count on finding a couple of issues.
Anyway this shouldn't be taken as a damnation of Java, which is worlds better than everything else I've used in both respects (although I have limited experience; for instance, I have never tried Smalltalk), just a note that it's not all sweetness and good in these respects.
Ok, I really like Java in general, and I'm not particularly a fan of C++ for all kinds of reasons that make me want to whack Stroustrup up side the head repeatedly. I've used both pretty much since their inceptions, including writing stupid amounts of code over the years.
Given care in writing the code, C++ pretty easily beats Java in both performance and memory footprint. Java has come a long way in terms of raw performance but it's still the case that in general you're not going to beat equivalent code written in C++. Memory footprint -- well, Java just sucks in terms of memory footprint. (I do realize there are some pretty tight implementations but that just eases the pain a bit, it doesn't get rid of it.)
Having said that I have found that Java developers are, on average, about three times as productive as C++, which means you can improve the application algorithmically three times as fast, and that pays big dividends over time. (It's hard to overstate how much I dislike the compile and link performance of C++, although I don't think this is the primary reason why Java is superior in this respect.) Java's garbage collector is likewise a big win in eliminating those annoying heap corruption problems that suck the life out of C++ development, and it's hard to overstate how good mandatory exception handling is at making applications robust in the face of failure. For example, the first commercial Java product I worked on never went down completely; this was a problem for the beta cycle because people often wouldn't consider "this button didn't do anything" to be serious enough to report, versus "the whole thing crashed" as is typical in C++.
So, I prefer Java when I can use it, or C# which is vastly superior for GUI development and fixes a number of really stupid Java design decisions (it's like he said in the prologue of Princess Bride, the book -- they kept "the good stuff"). But if memory is tight or if you really need high performance -- such as in gaming, real-time systems, or large-scale computation -- Java is pretty much a non-starter.
Replying to myself is probably bad form, but it occurs to me that Tomcat is similar, and JBoss too although the latter is a little harder for me to consider truly open source. For sure I'd rather use these products than WebSphere or WebLogic. I am not nearly so enamored with the whole development cycle versus Microsoft's tools though; Eclipse has a whole lot of strikes against it in my book.
Even so it seems to me, now that I'm thinking about it, that there are a lot of cases in web development where the open source stuff is pretty darn nice. Spring framework, for instance... although this lagged behind ATG's Dynamo, which has a startlingly similar design, by quite a few years.
Ahh, I've been pointing out that other commercial vs open source comparisons here are generally being done against inferior products, or the writer is just confused, but Apache is one case where I think the open source product is best-of-breed in so many ways.
IIS does have some things going for it, in particular it's a lot easier to set up and you gotta love the integration with the development tools, but Apache is certainly faster and more robust and isn't limited by the lousy performance of Windows and has a whole lot more extension options. There is a good reason why there really aren't any viable competitors on UNIX.
So... kudos, you identified a case where open source truly does beat the competition.
I'll second this. There are a lot of things to like about gcc, but highly performant code is not one of them. This is especially true once you get off of x86 targets.
Gcc was never intended to be a particularly good compiler in this respect, although it did beat out the AT&T cc quite handily when it was introduced... much to the surprise of Stallman, as I recall. Of course, that was a particularly stupid compiler and virtually all of the commercial UNIX providers replaced it with better stuff.
Having said that, I think it's the rare case these days where the difference matters (on the platforms where gcc is pretty good anyway, it's terrible on some of them) and gcc still sees active development whereas there doesn't seem to be a whole lot of effort going into commercial compiler development nowadays (with the notable exception of Microsoft).
Sourcesafe is horrible; it's not reasonable to take the worst of the commercial world and compare it to the best of the free software world. Instead, compare git to a decent commercial system, e.g. Perforce.
I note that I can't personally compare git to Perforce, although from quick Google checks it looks like the two are roughly comparable with Perforce winning out in some key areas (to me, at least) like native Windows support and GUI, to say nothing of documentation. In any case git is not clearly superior even technically according to numerous comparisons I found.
Other than cost, I'm curious as to what you are measuring.
It's not performance. Solaris 10 on a PC outperforms Linux when stressed, sometimes by huge margins. It's not a complete blow-out with modern kernels, like it was in the 2.4 days, but it's still significant -- and there are corners where it really still is a complete blow-out (like: What does fsync() do?).
It's not stability. Again, Solaris 10 is much less likely to crash or go all wonky when stressed, and AIX is similarly robust. And again there are corners here that can have large impact on reliability of applications (e.g. I recall some annoyances with direct io, although the specifics are escaping me at 3am).
It's not core features, by and large Linux lags the commercial UNIXen in that arena (e.g. Dtrace), which isn't really surprising since Linux is effectively a clone.
Having said that, there are certainly UNIXen that do not stand up so well to Linux (*cough* HP/UX *cough*), but there certainly exist UNIXen that are generally superior.
Of course, a lot depends on what you're doing with the system. If I need to push the system as hard as possible then I want Solaris if it is an option. If I am looking for general usability I find Linux preferable. The differences in the latter case are often minor (like the -c argument to "script") but taken en-masse they make for a more pleasant experience.
That's the way I see things today. I am not at all sure we won't see Linux pull into the lead over time -- it is getting better and better with age, and most UNIX systems appear to be seeing improvement stagnation. (And that may well be universal soon: What are the odds that Solaris doesn't see its R&D budget chopped under Oracle?)
In summary I would generally prefer to do development on Linux, but want production on Solaris. Best of both worlds, at least today. It may be telling, though, that on my personal server systems I run Linux even when I am concerned about performance -- the ongoing administration is significantly easier it you pick the right Linux distribution, and it isn't picky about hardware like Solaris is, and the tool suite is free as in beer without any extra effort.
Except on the desktop. There MacOS is supreme IMO. All the goodness of UNIX, plus stellar UI and commercial applications. I dunno how it does in stress situations (almost nobody buys Apple servers) but it's a great desktop system.
Since they don't give numbers for several years, but only compare this election to an election last year, I wonder whether or not last year's 44,000 number was due to large turnouts as a result of a presidential election cycle. I know that my town sees numbers an order of magnitude higher with such elections than the "norm", and in years that don't have either presidential or national representative races, like this one, the numbers are abysmal.
Not that I don't believe the mechanism they used hurt their numbers, just that it might not have been the only cause.
It still astounds me that twenty-plus years after BSD FFS we have NTFS doing absolutely nothing whatsoever to avoid initial fragmentation. You want to talk about a lazy monopoly, there's proof right there; nobody else still has this problem.
Back on the subject at hand, SSDs were always likely to have teething pains, that's one of the reasons I've avoided them. We saw the same kind of thing with compact flash, if you remember. Give 'em a couple more years to work through it and get the prices down and capacities up and laptops won't have spindles anymore.
Oh yea, and try hard not to use SSDs for swap. The very thought of a Windows paging storm to an SSD makes me think of a burning wallet.
I did mention that I thought the Thinkpads had similar characteristics to the Apples. A bit better initial reliability, really. But they aren't any less expensive. The Dells were trash in my experience; they always seemed like such good bang-for-the-buck but they just didn't last.
No personal experience with the Toughbooks but they aren't cheap either.
Even with good hardware I still find that ongoing maintenance of Windows makes it relatively expensive even if the hardware is in good shape.
Slashdot Mirror
This is not universally the case, of course; some of the books were about equally priced, excluding shipping, but of course there is always shipping. (I use Amazon Prime, so the shipping is not easy to calculate, but it's there.) I am a heavy reader -- 2 to 3 books a week. Over the first two years of Kindle ownership I saved more than $1200 versus paper if I'd purchased paper from Amazon. This is actual savings, not made up, I added them up in a spreadsheet in a fit of pique while arguing about e-book futures with someone. But really the savings were much greater: Many of my book buys are impulse, and that means I used to hit bookstores a lot and pay retail prices, especially for recent releases. (As an aside I lament the fact that e-books are the final nail in the coffin of local booksellers. I hate that, although I love having huge catalogs available all the time.)
Of course the readers ate into that a lot; $400 for the first one, $360 for the Kindle2 (because, what the heck, I saved more than that the first year anyway and my daughter can use the old one), and $200 for a refurb Kindle2 after I drove away with the first one on the trunk of my car (this when the readers were still $360). As of last summer I was really only about break-even, but of course every month I go without buying another reader is like another $50 so I'm well up again at this point (plus my daughter's books are cheaper too).
Now, those are all new book purchases, as is my norm. If you're one of those people who hits used bookstores or libraries the economics completely fall apart, although they are getting better as the reader prices drop.
Going forward the economics should only get better.
Dropping prices for e-ink readers are one reason I think the Kindle et al are pretty safe from the iPad. Most of the book readers I know weren't keen on spending $400 for a reader when the Kindle came out, though by last Christmas, at $260, many more made the jump. I think it's a safe bet that you'll see Kindle2-class readers for under $200 by the end of the year, and probably around $120 by the end of next year. It's going to be very hard for the iPad to compete on price. It's a different class of device, so perhaps it will do well anyway, but it isn't going to be mass-market in the way e-book readers are quickly becoming.
Personally I look forward to the competition in e-readers. The more of them that are out there the more competition from retailers and the stronger the incentive to standardize on one book format. (I bet we don't see DRM disappear entirely, for lots of reasons.)
Kind of awkward, sometimes. But we take a drumming for it on occasion, like when we confuse the two systems and our spacecraft don't work.
I was seriously concerned about this when I bought my first Kindle. You know what? Within 10 minutes you just don't see it anymore, pretty much the same way you blank out the page flip on a paper book. Perhaps this is because it really isn't a "flash".
It'd be ok with me if it went away, but it's not a problem or even a distraction.
They have -- in fact, E-Ink has been demonstrating the technology for four years:
http://eink.com/press/releases/pr86.html
and that's not the only similar tech. Wired had a summary of several back in June:
http://www.wired.com/gadgetlab/2009/06/blackandwhite_ebooks/
Commercial availability of E-Ink's color displays is expected in late 2010. I would bet pretty strongly that the Kindle 3 uses 'em.
I notice that the Nook does have a clip-on light, one of several places I think they are superior to the Kindle, but I haven't had the opportunity to see how well the unit works in practice and I'm not convinced that the LCD panel isn't a temporary hack until e-ink technology supports color. It certainly is an expensive feature in terms of battery life.
The shiny is just gravy; the important thing is that it works really well.
As an internet device anyway. Kinda so-so as a phone, although still worlds better than Windows Mobile. (I haven't figured out what Windows Mobile is actually good at, but perhaps the reason that it has such lousy market penetration is that no one else can, either.)
Considering the iphone as an e-book, it works reasonably well in a pinch. It's nowhere near as pleasant as a Kindle for long text, and I can't for the life of me figure out why everyone thinks swiping is a good way to turn pages when you're doing it every few seconds, and long reading stints drain the battery, but at least it fits in my pocket.
In any case many technologies will be employed in e-books over the next few years, you'll be able to pick and choose. I already use LCD and e-ink....
I realize that this is near-mandatory pain on Windows, alas, but there will have to be pretty significant threats before I would consider running such a thing on my phone.
What he said. A $200 service pack. Woo-hoo! Makes me want to buy more Windows boxes.
An Airport Extreme has about the same limit (that's what I'm using now because Apple Airport firmware updates kept breaking compatibility with the Linksys WiFi). In fact, I didn't find any consumer-grade stuff that did better than low to mid 30Mbps range last time I went looking (admittedly a year or two ago now).
I sure did find worse, though: Many of the Linksys WRT54G units could only pull around 14Mbps regardless of firmware. And for a little while I used a Netgear FVS318 but that was awful -- it peaked at about 8Mbps with NAT enabled! (It was what Staples had when the aforementioned WRT54GS decided to die at an inopportune time.) I do not recommend the Netgear; not only is it stupid slow, but if it sees heavy load you get frequent hangs. The only plus was that it was really easy to set up and was a lot better than not having any network.
Changing their behavior could make their use of Windows safer, but speaking from practical experience every time I have switched a problem user to a Mac (which in some cases I did at my own expense) the support time fell to essentially zero after a short adjustment period. Importantly, this has always happened. Changing the OS fixed the problem without fixing the user.
Theoretically the Mac (and Linux) can have similar malware problems to Windows, and there is proof-of-concept malware out there. Practically, though, it just isn't the case. I suspect that the security model -- which is much better than even Vista/Win7 when you get into the details -- has a lot to do with that. The claims that it's all about market share are wishful thinking; it's about ease of entry. It's very, very easy to subvert Windows ... and very difficult for Microsoft to fix this because a large amount of software depends on the ability to do things that should really not be allowed (like, say, injecting threads into other programs and using VirtualProtect to make code pages writeable -- with those capabilities you will never make a secure system). As long as it's much easier to subvert Windows there's not much point in going after the Mac.
But it goes beyond malware. Certain Windows features, such as the Registry, are hideously overcomplex and unreliable. They should have been tossed years ago -- and since everyone manages registry settings via APIs this could be done with excellent backwards compatibility (consider how easy it was for Apple to switch the format for Prefs in Snow Leopard). When something messes up the registry, all too common in my experience, there is little choice but to burn it down and rebuild from scratch. (Thank God for Acronis or I'd go insane from the reinstalls.)
If you want a system that doesn't require a lot of admin time I have to say that you can pick pretty much anything other than Windows and do well. It will be easier to set up, easier to back up, easier to fix if something goes wrong, and the software will be considerably less expensive (everyone else packages useful software in-the-box).
Of course, there can be overriding concerns that force the use of Windows ... and you take your lumps if that's the case.
Prior to the iPhone I rarely hit the web with my phone (it had to be an outright emergency). The plan made it stupid expensive and I got grey hairs waiting. But having seen how useful the iTouch was when it was on the net I went for an iPhone (AT&T be damned. No, really, damn AT&T). Having a useful web browser pretty much everywhere has been a godsend.
It's kind of a crappy phone, but it's a remarkably good browser despite the form factor. That should only get better across the industry as the devices and software gets more capable.
That free-as-in-beer and forced code contribution work hand in hand to benefit the end user is unquestionable, but I don't know many who make that an overriding (positive) concern for picking one product versus another.
(By the way, I hadn't known about this video and I appreciate the pointer; sometime when I have an hour to spare I'll watch the whole thing.)
I think cheap branching is more interesting; Perforce branching is too expensive IMO. (I note that there are commercial products that are good in this area, too, like AccuRev.)
The two places I'd be worried a lot about git are in merge management (how easy is conflict resolution?) and scalability. I know git has trouble with the latter, and I suspect it isn't all that great at the former. I have regularly been impressed with how well Perforce manages automatic merges with minimal conflicts that need to be resolved. In a highly active codebase it's hard to overstate the value of reducing the number of merge conflicts that have to be handled manually.
Aaanyway this was just a long-winded way of saying that I still don't think there is a clear winner.
I was going to argue that Linux' low cost was the key to its early success, but the more I thought about it the more I believed that it wasn't low cost alone that did it -- or even that it was the most free. It wasn't the most free, and still isn't; that crown goes to BSD no matter how you slice it, whose license is the next best thing to public domain.
I know Stallman likes to tout the "freedom" aspect of the code, but really the GPL works more like a traditional patent cross-licensing agreement, where parties share each others' technology because otherwise they wouldn't be able to ship a product at all, rather than being truly unlimited.
The GPL takes this to the extreme, forcing unlimited cross-licensing of any derivative work ... not only to a select few who can buy into the club, but anyone who has any desire at all to be involved.
This avoids the software version of the Tragedy of the Commons, which BSD has always seen: Lots of products use(d) BSD, but because everyone's changes were proprietary they were almost always kept in-house. BSD development therefore got little advantage from commercial development and the pace of its improvement has been slow to say the least. (It was further hampered, very significantly IMO, by a bad case of NIH in the core maintainers.)
The fact that commercial entities, who had the motivations and money to do major work, had to share their Linux improvements certainly led to major advances that have created a feedback cycle whose power is obvious at this point. This cycle continues not because the software is free as in freedom, but because it is not -- it comes at the cost of technology cross-licensing.
Hah, we always harped on Sun for that statement. We said it was "write once, debug everywhere." Not so much for server software, with a few notable exceptions where OS differences easily showed through the Java APIs, but UI development was certainly not "run anywhere." (Not that C++ was better by a long shot, but at least when you did it in C++ there was no expectation from your boss that it might actually work.)
As for "always backward compatible" ... you must be joking, or inexperienced. I don't think there was a single significant JDK release from 1995 to 2004, when I was focused almost exclusively on Java, where I didn't have to fix a bunch of stuff. Swing in particular tended to find new and irritating ways to not work with existing code. Again server code wasn't nearly as bad, but I think that's because it tended to touch a lot fewer core libraries. I don't recall anything that was difficult to fix, usually just tweaks here and there, but I assure you I got familiar with System.getProperty("java.version").
I see that one of the follow-on comments claims that this is due to poor coding, but I didn't find that to be the case. I saw behavioral differences in libraries that were sometimes inexplicable, although more often it was that they fixed bugs in the libraries. There were a lot of bugs outside of the core classes, especially in stuff like Swing and JavaMail. I never had to write a lot of version-specific code, but it certainly happened often enough that you couldn't haphazardly take an upgrade of the runtime system. You could pretty much count on finding a couple of issues.
Anyway this shouldn't be taken as a damnation of Java, which is worlds better than everything else I've used in both respects (although I have limited experience; for instance, I have never tried Smalltalk), just a note that it's not all sweetness and good in these respects.
Given care in writing the code, C++ pretty easily beats Java in both performance and memory footprint. Java has come a long way in terms of raw performance but it's still the case that in general you're not going to beat equivalent code written in C++. Memory footprint -- well, Java just sucks in terms of memory footprint. (I do realize there are some pretty tight implementations but that just eases the pain a bit, it doesn't get rid of it.)
Having said that I have found that Java developers are, on average, about three times as productive as C++, which means you can improve the application algorithmically three times as fast, and that pays big dividends over time. (It's hard to overstate how much I dislike the compile and link performance of C++, although I don't think this is the primary reason why Java is superior in this respect.) Java's garbage collector is likewise a big win in eliminating those annoying heap corruption problems that suck the life out of C++ development, and it's hard to overstate how good mandatory exception handling is at making applications robust in the face of failure. For example, the first commercial Java product I worked on never went down completely; this was a problem for the beta cycle because people often wouldn't consider "this button didn't do anything" to be serious enough to report, versus "the whole thing crashed" as is typical in C++.
So, I prefer Java when I can use it, or C# which is vastly superior for GUI development and fixes a number of really stupid Java design decisions (it's like he said in the prologue of Princess Bride, the book -- they kept "the good stuff"). But if memory is tight or if you really need high performance -- such as in gaming, real-time systems, or large-scale computation -- Java is pretty much a non-starter.
Even so it seems to me, now that I'm thinking about it, that there are a lot of cases in web development where the open source stuff is pretty darn nice. Spring framework, for instance ... although this lagged behind ATG's Dynamo, which has a startlingly similar design, by quite a few years.
IIS does have some things going for it, in particular it's a lot easier to set up and you gotta love the integration with the development tools, but Apache is certainly faster and more robust and isn't limited by the lousy performance of Windows and has a whole lot more extension options. There is a good reason why there really aren't any viable competitors on UNIX.
So ... kudos, you identified a case where open source truly does beat the competition.
Gcc was never intended to be a particularly good compiler in this respect, although it did beat out the AT&T cc quite handily when it was introduced ... much to the surprise of Stallman, as I recall. Of course, that was a particularly stupid compiler and virtually all of the commercial UNIX providers replaced it with better stuff.
Having said that, I think it's the rare case these days where the difference matters (on the platforms where gcc is pretty good anyway, it's terrible on some of them) and gcc still sees active development whereas there doesn't seem to be a whole lot of effort going into commercial compiler development nowadays (with the notable exception of Microsoft).
I note that I can't personally compare git to Perforce, although from quick Google checks it looks like the two are roughly comparable with Perforce winning out in some key areas (to me, at least) like native Windows support and GUI, to say nothing of documentation. In any case git is not clearly superior even technically according to numerous comparisons I found.
Other than cost, I'm curious as to what you are measuring.
It's not performance. Solaris 10 on a PC outperforms Linux when stressed, sometimes by huge margins. It's not a complete blow-out with modern kernels, like it was in the 2.4 days, but it's still significant -- and there are corners where it really still is a complete blow-out (like: What does fsync() do?).
It's not stability. Again, Solaris 10 is much less likely to crash or go all wonky when stressed, and AIX is similarly robust. And again there are corners here that can have large impact on reliability of applications (e.g. I recall some annoyances with direct io, although the specifics are escaping me at 3am).
It's not core features, by and large Linux lags the commercial UNIXen in that arena (e.g. Dtrace), which isn't really surprising since Linux is effectively a clone.
Having said that, there are certainly UNIXen that do not stand up so well to Linux (*cough* HP/UX *cough*), but there certainly exist UNIXen that are generally superior.
Of course, a lot depends on what you're doing with the system. If I need to push the system as hard as possible then I want Solaris if it is an option. If I am looking for general usability I find Linux preferable. The differences in the latter case are often minor (like the -c argument to "script") but taken en-masse they make for a more pleasant experience.
That's the way I see things today. I am not at all sure we won't see Linux pull into the lead over time -- it is getting better and better with age, and most UNIX systems appear to be seeing improvement stagnation. (And that may well be universal soon: What are the odds that Solaris doesn't see its R&D budget chopped under Oracle?) In summary I would generally prefer to do development on Linux, but want production on Solaris. Best of both worlds, at least today. It may be telling, though, that on my personal server systems I run Linux even when I am concerned about performance -- the ongoing administration is significantly easier it you pick the right Linux distribution, and it isn't picky about hardware like Solaris is, and the tool suite is free as in beer without any extra effort.
Except on the desktop. There MacOS is supreme IMO. All the goodness of UNIX, plus stellar UI and commercial applications. I dunno how it does in stress situations (almost nobody buys Apple servers) but it's a great desktop system.
Not that I don't believe the mechanism they used hurt their numbers, just that it might not have been the only cause.
It still astounds me that twenty-plus years after BSD FFS we have NTFS doing absolutely nothing whatsoever to avoid initial fragmentation. You want to talk about a lazy monopoly, there's proof right there; nobody else still has this problem.
Back on the subject at hand, SSDs were always likely to have teething pains, that's one of the reasons I've avoided them. We saw the same kind of thing with compact flash, if you remember. Give 'em a couple more years to work through it and get the prices down and capacities up and laptops won't have spindles anymore.
Oh yea, and try hard not to use SSDs for swap. The very thought of a Windows paging storm to an SSD makes me think of a burning wallet.
No personal experience with the Toughbooks but they aren't cheap either.
Even with good hardware I still find that ongoing maintenance of Windows makes it relatively expensive even if the hardware is in good shape.