How Software Companies Die (1995)

Really? This is like a nerd mantra, it hits so many stereotypes in so few words. Orson talked about it at BayCon once when it came up.

I don't have a clear memory of exactly what he said but I took away from it that smart, introverted, people have a terrible time seeing anything outside their pool of influence, like looking at your shoes in a crowd and saying "everyone is crowding around me because I'm the coolest" when if you looked up you would see a pack of wolves circling the crowd. So all of their evaluation in their environment is about them and how people around them are responding.

An artifact of that inward looking view is to construct an internally consistent explanation for why things suck now but didn't before. That always targets people who are not like them (the introvert) and for whom the introvert cannot understand their motivations or values. (and more often than not in a software company that is "suits", aka business people or managers)

You combine that with a changing market place, and the thing that made the software successful before isn't valid any more, but if the programmers don't know why the software was successful in the first place they can't understand what changed or how to fix it. Imagine an artist who paints with charcoal, gets discovered, people buy their works, and then fashion moves on, and the artist goes broke. To an external viewer, fashion simply moved on, the artist was there (and fashionable for a time) and then they weren't. But internally the artist may have no concept of how briefly they were fashionable and now they weren't. Instead they construct an externalization of "discovery" where suddenly people begin to understand what it is they are doing with their art, and when they move on it is always some external force which has corrupted the minds of people seeing it or modified their vision. To accept the alternative, that their success was luck and fashion, rather than some deeper meaning, may be too harsh to think.

People cling to sense of destiny and purpose rather than consider themselves the random connection of ova and zygote and circumstance. Where is the meaning in that?

Orson captured this tension wonderfully in his essay, it resonates both with the bees who desperately want to be special and with the bee keepers who want productive bees without understanding that the bee's production is a byproduct of flowers and normal bee activity, it's a side effect of where the hive is and where the flowers are. Many software companies are fashion objects, successful ones are constantly producing new concepts and new ideas to stay in the limelight. Unsuccessful ones take a single success and assume they are done, nothing left to do but iterate. But fashion doesn't work like that.

I continued it with my (ongoing) series on my blog (8 parts and counting, will take a while to read): http://michaelochurch.wordpress.com/2013/02/19/gervais-princ...

Companies start out as a risk transfer. Those who can tolerate financial volatility and change and social upheaval (MacLeod Sociopaths) start to trade risk with those who don't want it (MacLeod Losers). The risk-seekers become the entrepreneurs; the risk-averse become workers. This is benign for a while, but as the company becomes stable, the upper class becomes entitled, established, and complacent. They want to keep the perceived risk trade going, but the business is actually de-risked, so now they're just grabbing for unjustified or legacy superiority. They've won, and now they want to stay where they are. In order to keep the risk-transfer between them and the (expectancy-sacrificing) Losers, they have to create a middling buffer class of Clueless. This style of organization becomes resistant to change, systemically underperforming, and bland, but it's extremely stable on its own terms. People don't want the jobs above them enough to break the rules. Losers (who value comfort and stability) see the ranks above them as being harder, riskier, and are happy to stay put. Clueless are oblivious to the Effort Thermocline (the point in an organization where jobs become easier-- but also more political-- with ascending rank instead of harder) and run the organizational day-to-day while oblivious to the agendas of the Sociopaths above them.

The most common end state of this is the organization with the full MacLeod hierarchy: Losers at the bottom, Clueless in the middle, Sociopaths at the top. It's a mature rank culture where position and subordinacy matter more than creative excellence and technical improvement. But why?

The process is complex, but the truth is that this form of organizational corruption was, for a long time, harmless from a business perspective. When you're providing commodity products or services, the MacLeod rank culture works. It does the job.

What's changing (first in software, increasingly in other areas of business) is the convexity switch. Concave work is work on which the difference between excellence and mediocrity is insignificant compared to that between mediocrity and zero. It's "get-it-done" commodity work where creativity isn't necessary or even useful. Convex work is that in which the difference between excellence and mediocrity is huge, and mediocrity is close to (or might be) zero.

The difference between the (winding down) industrial and (soon to come) technological eras is the convexity of the work that the world needs. In the industrial era, most needed work was concave and the hierarchical MacLeod organization worked well. In the technological era in which the commodity work has been given to machines (who do it far more reliably and cheaply than any human) what's left is convex work. Traditional industrial management, though successful for 200 years, just falls down. The reason why is that it's all about reducing variance. For concave work, variation is mostly to the downside, and variance-reduction (the goal of most management) improves expectancy. For concave work, variance is mostly to the upside, and variance-reducing management drives expectancy to zero.