I know many Linux and open source advocates want to claim that they have replacements for every conceivable software package. Yesterday, I dealt with two separate issues that prove that no matter how much improvement there’s been in open source software, the benchmarks against which they’re compared are the proprietary applications they seek to mimic; and open source rarely achieves parity with its proprietary counterparts.
One of the problems with open source is too many people want to reinvent wheels rather than join ongoing development of existing projects. So instead of all the time and resources going into making one thing work better, we end up with several things that don’t work; and more often than not, lists of things “to do” remain longer than lists of stable features. And also more often than not, the comparisons that any way relate to one application or utility being a “clone” are laughable because the proprietary versions aren’t sitting targets and because they have far more working, stable features.
Among the poor comparisons are OOo for Microsoft Office, GIMP for Photoshop, Gnash for Flash, and xsane for any proprietary scanning utility (which typically comes “free” with the scanner/printer). While these projects have vastly improved, they’re not really in the same league as the proprietary applications they seek to replace.
I take news of forks and “new” projects with a bit of hesitation, except when they’re backed by big money — that means companies with a vested financial interest in open source. It’s the people from Novell, IBM, Red Hat, and so on — and the money those companies have put into development — who’ve made the greatest improvements in everything from Gnome to NetworkManager to the Linux kernel to a lot of the little things that make Linux approachable by a lot more people than it otherwise would.
I saw this morning courtesy of Lifehacker that the Ubuntards (Canonical) are reinventing another wheel. It’s scanning this time.
Although his software isn’t officially at a finished, 1.0 stage, it’s already decent enough to be an attractive install…
Most open source software isn’t at “a finished, 1.0 stage.”And I’m not sure what the writer at Lifehacker considers 1.0 — I looked at the bugs listed at the launchpad site for simplescan and can say right off the bat that it’s not even close to being as functional as the much more mature xsane. And with the scanner I use most often, I can say that xsane isn’t nearly as functional as the Windows utility that came with my scanner .
The most recent blueprint notes one of the issues facing development of such a program:
A general problem is to ensure that white is really white (your interface mockup displays a scanned page with gray background). A simple scan interface does not serve so much if the scans display gray backgrounds, and necessitate to open gimp to pick the white point to correct balance. User could check “automatic white balance” box in case of scanned pages with white background.
In fact, among the “bugs” listed as wishlist items are things I take for granted using Windows-based utilities — including real greyscale (as opposed to black-white) and color separation.
I’m not knocking this project or even Canonical; I know that Canonical’s philosophy is to make everything work simply even for the least savvy computer user. It just seems that a lot of the money and (wo)manpower these projects throw at reinventing wheels could go a long ways to bringing existing projects up to speed so they’re legitimate contenders against proprietary counterparts.
The wheel doesn’t need constant reinventing. Right now, most open source projects only compete against other open source projects rather than against proprietary software. For some things, this is probably just fine. In the bigger picture, though, it leads to stark choices between half-implemented, amateurish efforts from the open source world and well-polished proprietary applications that work as intended. It would be nice if open source advocates would improve what they already have so that it’s stable and works well. Instead they’re constantly shifting goalposts with changes in APIs and spending more time with paradigm shifts in user interfaces. Broken support, broken applications. When that happens with Microsoft or Apple, they’re held accountable by consumers — screw up enough times and people will stop buying your software. When that happens in open source, the user is left holding the bag until the user or developers, anyone, makes a fix.
I’ve dealt with this in several ways. I bought a device that was ogg-friendly but found out MTP devices lack open source support, so using it under Linux and BSD has been very dicey — but the damn thing works great under Windows and syncs through WMP like a charm. The same trouble has proven true with onboard hardware, including the card readers in my AA1 (flawless in Windows, only one works in Linux and a card has to be inserted before booting or else it won’t ever read a card), wireless (I quit using Linux altogether on the AA1 due to rampant, incorrigiable issues with ath5k), etc.
Some day maybe all this stuff will work as well under open source as it does currently under proprietary software. Until then, proprietary software isn’t evil: it fills a valid niche for people whose computers and hardware have to work in a certain and consistent manner. But as long as open source continues forking without good cause and duplicating its own labor over the same recurring sets of issues, it’s only going to improve slowly. And the slower that ever-reinvented wheel turns, the faster it’s going to get left behind. And why — just for some turf battles over who controls any particular project, not over anything substantive. That’s petty, and it’s also evil.