% fortune -ae paul murphy

Alternate Universes: Unix vs MCSE

Just so you know who to blame, here's a small part of a comment from last week:

I read most posts on this blog, and to be honest, I find it hard to agree with a lot of what you say. Sometimes I wonder if you live in an alternate universe to mine, linked by some freak blog worm hole.

...

TheTruthisOutThere

One of the things that happens when people first adopt and then adapt specific toolsets is that the technologies and operational methods start to co-evolve and, as that process gains momentum, the combination becomes increasingly resistant to external information.

One of the consequences of this is that people working within such a process can develop a view of their world at work that looks completely consistent to them and yet is totally at variance with what's going on in the world outside - and because multiple technologies and markets can evolve concurrently, there can be many such behavioral complexes whose members can interact - and it's the resultant communications difficulties that give rise to the "alternate universes" agent Truth sees.

Thus when I talk about "desktop reliability" my Solaris/Sun Ray biases mean that I'm thinking in terms of install and forget systems that run for years without change or significant attention - but the Windows guy I'm talking to hears the word in the context of PC security, weekly software change, and an eighteen month hardware replacement cycle.

Similarly when I talk about small document files I'm usually thinking in terms of a few K; but he's hearing me in terms of megabyte sized Word files - and when I mention "terabytes" in a data storage context I'm typically thinking about unique data plus overheads but he hears me in terms of hundreds, even thousands, of replications of the same few gigabytes.

I believe that the most important operational consequence this has for management is simply that failing to match cultures to technologies is a recipe for utter failure. On the world stage this is why generals brilliantly fighting the last war usually kill thousands of their own soldiers and on the IT scale it's why putting a data processing expert in charge of Unix is akin to dropping an 1890s steam engineer into the pilots seat of Boeing 787.

The problem is that a computer isn't just a computer: it's the visible element in an evolved behavioral system. Thus within context consistency makes sense - i.e. putting a guy with thirty years of mainframe experience in charge of a data processing shop makes sense because his reactions and unquestioned beliefs will be consistent with the work you ask him to do. Taking him out of context, however, does not: put the same guy in charge of a Solaris system and the guy will grow increasingly frustrated and unhappy while user complaints, costs, and system failures all sky rocket - because all of the things he knows for sure about his job, and usually doesn't even know he acts on, evolved to fit the 1920s data processing heritage he knew in the IBM environment but are dead wrong for a science based systems environment.

The problem isn't the people: that MCSE who interprets the word "reliability" in the context of the technology keeping him employed is neither wrong nor bad; the problem is that the certainties which co-evolved with his technology didn't evolve in the same directions in mine - meaning that we use the same words to refer to very different concepts and think we're communicating when we're not.

If he and I, for example, were to tour a former client still running 230 NCD HMX terminals I helped install in 1990, he'd probably see an IT horror story where I see a system that's so successful the users never give it a thought.

(The story on the one continuity failure they've had is hilarious. Apparently the procurement people procured movers whose primary qualification consisted of being from Quebec and therefore refusing to acknowledge English as a working language. Anyway, they'd moved about half the people and HMXs to the new offices (taking advantage of the move to replace the original 21" Hitachi CRTs with 20" LCDs) and were in application upgrade test on the new primary processor when the movers trundled a forklift into the original Annex (really just a bunch of trailers) and cheerfully made off with the pallet holding the battery pack and generators running the operational processor - apparently trailing cables, sparks, diesel fuel, and the occasional Lantronix switch.)

So, bottom line, does all this mean truthseeker's "alternate universes" comment is right? I think so - as long as we all understand that the barriers to cross cultural communications are natural consequences of differences in technical and historical context, and thus reflect neither ill-will nor incompetence on the part of the people involved.


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.