Exploration Through Example

A transcript of an OOPSLA talk: Table of contents

A post by Clarke Ching finally nudged me to finish this Latour series. Here’s the key excerpt:

The bad news is that I think we are about to fuck it all up because (according to my jaded view of the world) we - the agile community - are focusing on the wrong things. We are focusing on optimising our technical skills (the bits we are naturally good at), but we are ignoring the skills we really need to foster within our community. What are these skills? I’d like to call them marketing and sales skills, but I think they’ll be better received if I call them “influencing” skills.

I’d like to disagree in the sense that I think things are even worse. Latour shows how.

Here are two statements that drive me crazy:

A good manager can manage anything.

The net treats censorship as damage and routes around it.

Neither of them is true. Start with the first. Remember the baboons?

thirteenmonths.com

That baboon manager is ignoring how important things are for organizing and maintaining social structure. Someone who thinks a good manager can manage anything thinks that individual relationships are everything; things are nothing. Oddly, those people are often extremely good at individual relationships.

The net enthusiast is making the opposite mistake: things are everything; the social is nothing. She’s assigning one specific technology (packet switching) greater power than a complex social structure acting to prevent people from seeing information or, if they do see it, encouraging them to dismiss it. (It’s a mistake to think of censorship as an isolated act.) She’d find the subtle process by which Pasteur “converted” Anthrax from a disease to a bacillus uninteresting: since Anthrax is caused by a bacillus, Pasteur’s social network (comprised of thousands of people who had very specific effects) was inevitable.

I claim that, especially in the early days of Agile, there was a semi-conscious ANT-esque mixture between things and people and social structure and networks of influence.

Consider JUnit and the early practice of test-driven design. It encouraged pair programming because it’s harder to cheat and not write tests when you and your pair have to agree. (Pairing turns a private vice into a social event.) JUnit and pair programming both encouraged and enabled collective code ownership. “Encouraged” because collective code ownership reveals missing tests; “enabled” because tests make it safer to change someone else’s code. JUnit, TDD, pairing, and collective code ownership encouraged the development of continuous integration, which is itself an amalgam of tools, techniques, and social structures (such as those created at conferences).

JUnit, with its associated things and technical actions, helps build the team as a social unit. Early Agile, especially XP, had a bias toward building a network to bind the social and technical together.

To a large extent, I think that’s been lost. Clarke correctly points to technologists retreating to the safety of the most purely technical practices. (Kent Beck made the same point when he responded to Glenn Vanderburg’s OOPSLA talk on the relationship among XP practices.)

However, there’s been an equal and opposite retreat on the part of the humanists. People who talk about team-building are extremely unlikely to talk about JUnit’s role. Instead, team building is (if you’re a Scrum Master) about tactics like keeping your mouth shut and forcing people to self-organize. If you’re a regular attendee of AYE, it’s more likely to be about helping people to become congruent, using Satir-derived techniques that model teams as dysfunctional families.

No actual things with a pedigree more recent than the talking stick need apply.

Both retreats are alarming. Because of my personality, I’m more alarmed by the renewed ascendency of humanism and managerialism than Clarke is. I point to the website of the Agile Project Leadership Network:

This is absolutely standard leadership iconography: the hero, alone, struggling in a forbidding or at least exceptional environment. Anyone who thinks such iconography doesn’t matter probably also thinks that advertising doesn’t affect her and that she really reacts to $99.99 as being one cent less than $100.00. I think it works to maintain the status quo by reinforcing the self-image of the people who need to change as much as team members do.

I think we need to get back to our roots and try to recapture some of the strangeness of Agile. Think about it. Pair programming? What a stupidly inefficient idea! Bullpens? How could anyone think? Insisting on delivering something in a month? We can’t even order the servers in that time! Deliberately deciding to be reactive instead of proactive? But haven’t we spent our lives being told being proactive is better?

All that stuff was crazy. We need another generation of crazed thinkers constructing new actor-networks, new mixtures of the social and the technical.