% fortune -ae paul murphy

The most exciting thing in IT

Is a product that does exactly what it's supposed to and nothing else.

There aren't many - I've worked with hundreds, and played with thousands, of computer products and the list of things that stand out in memory as having genuinely met expectation is shockingly small:

  1. BSD 4.3 on Vax and MC68000 captured the loyalties of thousands of academic researchers and programmers including me. Why? because anything it was supposed to be able to do, it did - and no surprises.

  2. Phil van Cleave's APL68000 had a few minor bugs but the later releases worked flawlessly. Again: anything it was supposed to be able to do, it did, and nothing else.

  3. Adobe FrameMaker through to release 6.0 worked as intended. Frame is hard for people who think of Word as a document preparation tool to learn to use, but is, or at least was, a product you could trust because anything it was supposed to be able to do, it did - and nothing else.

  4. Vsifax - at least until they went after the PC market, was a scriptable, command line, package that did everything it was supposed to and absolutely nothing else. Wonderful.

  5. Unify's Vision Accell/U2000 combination had bugs, but minor ones and exemplified what a rapid prototyping environment should be - as unstoppable as Oracle 7.3 with Oracle Forms but with broader scope and much lower costs.

And beyond those, there's what? Perl, Apache, and Solaris 10 with ZFS are candidates, but then there's a big gap before you get to things like SuSe 7.1 and maybe the only Microsoft product worth mentioning: Excel 4.0.

There's even less of obvious excellence on the hardware side: HP's K series was insanely great for its time, you could throw the original AS/400s off a moving truck in the middle of rainstorm and as long as you schleped out enough batteries with them the things wouldn't even hiccup, early QMS printers were years ahead on software and Marine tough, NCD's HMX and 19C X-terminals were 100% reliable over decades, Sun's SPARC and UltraSPARC hardware (except for the PCs with SPARC chips like the Ultra5s ) has consistently been engineered for extended reliability - and then there's Lantronix: for a long time the best of the best, the TRW of network connectivity: near milspec stuff with software that worked exactly as advertised.

And that's about it. Think about that: tens of thousands of market leading but crappy products: from x86 to BASIC, thousands of good enough products: from IBM's 4381 series to the last Mac G4 laptops, from System V to the Java virtual machine - but if you set as your criteria that the product do exactly what it's supposed to do and absolutely nothing else under all reasonable operating conditions, I don't think your list of real winners will be much longer than mine.

Paul Murphy wrote and published The Unix Guide to Defenestration. Murphy is a 25-year veteran of the I.T. consulting industry, specializing in Unix and Unix-related management issues.