On Jan 7, 2015, at 7:02 AM, Les Mikesell <lesmikesell at gmail.com> wrote: > There's still a very odd mix of art and science involved. Yes. This is part of what I was getting at with my definition of “technology.” Once a thing becomes reliable, it stops being technology. It’s been reduced to the point Mr. Always Learning wants, something that merely serves as a tool to improve our lives. CentOS is far from perfect, however; it’s definitely still “technology.” That means it’s going to change. > still it seems like when everyone has the same > problem from the same causes there would be some way to automate or > re-use the knowledge of the fix instead of making everyone spend time > on their own new creative version. Let’s reflect on two maxims often heard in computing: 1. Don’t reinvent the wheel. 2. To one who only has a hammer, every problem looks like a nail. Both capture a slice of truth, but there’s a tension between them. If the hammer is solving the problem, why replace it? Because it isn’t the right tool for the job. What if an better tool exists, but it still isn’t the *right* tool? Which text editor should we throw out of the distribution as redundant: emacs, vi, nano, joe, or gedit? This is one of the consequences of openness. It’s why Debian has over 100 forks. Try this concept on for size: creative destruction. Sometimes you have to tear something down in order to build something better in its place. GNOME, KDE, XFCE, or OpenBox? Would the world really be a better place if CDE had never been replaced? Me, I’ll take GNOME 3 and all its warts over CDE any day of the week. CDE never would have *evolved* to be the equal of GNOME; it had to be destroyed to make room. Perl, Python, Ruby, C, C++, or Java? A whole lot of brilliance has gone into replacing FORTRAN, COBOL and Lisp, but is the current zoo of high-level languages a net improvement? As a programmer, I think so, but I’ll bet we can find someone who will say that brilliance would have been better redirected to other pursuits, even if it meant continuing to program in those languages. These people usually aren’t programmers, though, so I’m not inclined to give their opinions much weight. PostgreSQL, MySQL, or SQLite? How many DBMSes do we really need? SQLite simply cannot replace a lot of PostgreSQL installations. Likewise, it’s usually a blunder to try to use PostgreSQL in places where SQLite is most often found. (I’ve only seen it tried once; it was indeed a blunder.) MySQL can stretch to touch either extreme, but it’s not best used at these extremes. Subversion, Git, or Fossil? Do we really need all the version control systems we have? Can’t we all just agree on git and call it “done”? Again, programmer here: I’m glad we have a choice. The three I listed fit into very different niches. I’m glad we’ve destroyed CVS and SourceSafe; the world is a better place now. The time it’s taken people to convert to something better doesn’t bother me in the slightest. I gladly spent some of that time myself. Trump card: Linux. Need I say more? That’s creative destruction writ large.