Pivots don't start enough businesses

Probably no newsletter for the next few weeks — I'd like to get the thing-I'm-working-on into shareable shape by mid-September. I keep starting Mondays by thinking, "oh I'll get a post out in an hour or two and then spent the rest of my day coding" and then it's 4pm and I'm three drafts in. I'm going to put those hours into coding to start with, for at least the next few weeks.

Lots of feedback — comments and questions — on my post about team allocation last week, which is great, but as is often the case I wrote about a thousand words and now it’s halfway through the afternoon and I still don’t have a newsletter I’m quit happy with.

(More and more I think the Pivotal stuff is turning into a book.)

Instead of addressing any of that directly, I’m going to try to convince you to start a company — especially if you’re a Pivot.

I don’t think Pivots start enough businesses.

I think a lot about what “the ex-Pivotal community needs” and “what the Extreme Programming community needs.” One of the big ones is “more companies that work along these lines” — pairing, tests, group code ownership, etc. I get asked a fair bit things like, “How do I get a job with pairing?” and I don’t have a lot to tell these folks — right now the supply of jobs with pairing etc. is much smaller than the demand for them. This is partly a function of the way the software job market is generally, but it’s also a symptom of a problem:

Pivots don’t start businesses. And when they do start businesses, they usually start consultancies. And when they don’t start consultancies, they don’t use Extreme Programming — they say things like “XP doesn’t really work for startups.”

Instead, Pivots tend to join other companies and try to transform them — or to start consulting companies that try to transform businesses. Quite natural, given both the origin of the company as a consulting organization, and the economics of consultancies. Consultancies have very low startup costs, and require developing fewer skills than other kinds of companies — take basically any technical skill and add “enterprise sales” and bam— consultancy. And if you’re reading this you can already do at least a bit of enterprise sales, since you need to do it to get a W-2 job. Making a living from 1099s requires only slightly more.

But the more I think about “what made Pivotal the way it was” and “why aren’t there more companies like Pivotal,” the more that I think transformation is a bit of a dead end. Not useless — if you love transformation and it makes you happy, keep doing it — but at a large scale I think we’ve wrung most of what we’re going to get out of transformation.

Take a look at where we are now, as an industry. Most companies write at least some tests and wish they’d write more. Most companies do continuous deployment and integrate — if not continuously then certainly frequently. Most people pair at least occasionally — the outright “that could never work” resistance I remember encountering all the time in the earlier ‘10s seems rare — I think largely because remote work has made it much easier for both members of a pair to have their own keyboards.

I think transformation oriented-consultancies had something to do with this but mostly I think this happened because — and this is oversimplifying a bit but bear with me — Google showed up, started catapulting code into the cloud 500 times a day, made everyone who said that was impossible look really stupid, and at the same time trained tens of thousands of engineers who then often went on to start their own companies.

It’s very hard to start a company — this is the main reason more people don’t do it. But if you want to work in a very specific way I think it is vastly easier to start one from scratch than it is to take an existing company that has had some measure of success and get them to do things another way.