last chapters and how to avoid them
a review of "Recoding America", by Jennifer Pahlka
I have a bit of a reputation when it comes to not reviewing books that I ought to review – the trouble is mainly that I am easily bored, so the only books that grab me are ones that have a load of ideas in them. This makes it difficult to write a review that isn’t far too long to read, demoralises me when it comes to starting the project, and means that I tend to have half a dozen half-completed reviews on my laptop, all of them awaiting some idea of how to write a proper conclusion.
But this morning I came up with a genius solution to the last of these three problems, so here we go with “Recoding America: Why Government is Failing in the Digital Age and How We Can Do Better” by Jennifer Palkha. Strong positive recommend. It’s kind of under-served by its title – I read it about a year too late because people kept recommending it to me and I thought “hmm computer stuff”. But it’s actually a very good and very general book about management, government and structure – it’s rather like “The Mythical Man Month” in that it’s only really about technology because that happens to be the industry the author worked in. (Specifically, the book details her experiences as the founder of Code for America, and as the deputy Chief Technology Officer in the Obama administration).
It is pretty punchy to compare something to one of the acknowledged classics of management literature of the last century, but honestly, I think Recoding America is literally that good. Like Fred Brooks’ masterpiece, it’s really about human beings, communication and complexity. As I occasionally say, some things are big because they’re complicated, some things are complicated because they’re big and some things are both.
Websites, for example, are pretty simple when they’re small, but start getting very complicated indeed when they have to serve lots and lots of people. They’re complicated because they’re big. Healthcare administration is for the most part big because it’s complicated – there are loads of special cases that need to be taken into account and things are changing all the time. Consequently, something like the website for the US Veterans’ Administration was always at risk of turning into an organisational disaster, and it does indeed form one of the main case studies.
Anyone who’s read a bit of cybernetics will get the very strong sense of something I’ve written about on this ‘stack on numerous occasions in the past – that left to themselves with an important and difficult organisational task to solve, intelligent engineers almost always reinvent things which are recognisably the Principle of Sufficient Variety, the concept of homeostasis, the need for connections between system and metasystem, and so on. They’re all in the book, particularly the clear observation that the government IT procurement system (and implicitly, lots of other systems) works according to a set of rules which simply don’t have sufficient variety to cope with the system they’re meant to regulate.
Consequently, the rules get used as accountability sinks, and you get a sort of ecosystem building up in which people might complain from time to time about the overwhelming sense of sclerosis, but in which everyone’s daily living is dependent on operating within the rules, rather than adjusting the rules to fit the situation. It would be easy to write a book like that which was really annoying and just full of zingers against bureaucrats, but one of the impressive things about Recoding America is that it doesn’t do this; there’s a real empathy and attempt to understand why people do what they do, even when it appears objectively crazy and counterproductive. The basic problem is that everyone’s too busy chopping wood to sharpen the saw, and if I were to try to summarise the message of the book, it’s that the nature of complex systems management that small investments in higher-level coordinating and intelligence functions can have utterly disproportionate effect compared to much larger expenditure “at the coal face”.
But what I really like about “Recoding America” is that it doesn’t suffer from the great curse of nonfiction writers, “Crap Last Chapter Syndrome”. As anyone who has read a big thinky book (yes yes, including both of mine) knows, there’s an inevitable nose-dive in quality when, after a dozen chapters dealing the subtleties and origins of a complex phenomenon or wicked problem, the author is meant to sum everything up and offer a succinct and glib policy solution. I don’t know why this is the convention – I can only assume that readers like it so you have yourselves to blame. “Recoding America” gets over this by, more or less, making every chapter a last chapter, detailing the actual solutions that were found to specific problems, or the reasons why things never really got satisfactorily resolved.
Which is a technique I think I’m going to try to steal; if you go through enough of these specific cases, there’s much less pressure to pretend to discover general silver-bullet principles. Which, if you’ve taken on a problem that’s worth writing a book about, and treating that problem with sufficient respect, is definitely something you shouldn’t be doing.
Anyway, check out the book. It reads extremely well – I’ve described it to others as “rattling along like a Jack Reacher novel”, in that every page has something happening on it that moves the action along. And that is the end of my review of “Recoding America”.
Interesting. Worked as a contractor for the Social Security Administration under Obama and I noted a few things.
First off, "sclerotic" doesn't break. Much govie system behavior can be understood in terms of water delivery: suitably clean tap water is available in a Six Sigma level of quality *all the time everywhere*. (When it's not, as in Flint, it makes national news.) Similarly, govie systems typically have a small handful of clean-water-delivery equivalents. At SSA, it was making sure everyone in the US got their check on time. (By the way, this task is at least 3x as difficult as you think it might be. Two words: addresses and languages.) A secondary objective was Whatever the President Said. This varied from administration to administration, of course, but generally not *too* much, and the President knew he could only Say a very few things. (When I was there: speed up disability status determination. It happened, and it was a sophisticated technical endeavor not normally associated with "inefficient government.")
I was a solvent, really, as a contractor. I had a higher level of skill and education than my full-time govie colleagues, but I quickly realized they were far better at maintaining and piloting the system, even with their imperfect local views. Who to listen to, who to ignore, who to obey IMMEDIATELY, who you *actually* needed to talk to in order to do something, etc. My job was to unstick things--to use a military metaphor, I was Special Forces or a mercenary, doing jobs too sensitive or messy for regular Army, but those guys don't win wars and they certainly don't hold territory. If you put them in charge, they tend to break stuff by, say, getting rid of all doors in a country so they no longer have to waste energy kicking them down. That might be OK for a building (a single project), but it doesn't work at scale.
Most people who want to "make government more efficient" are entirely missing the point. The best way to make something more efficient is to fire difficult customers. Those are most citizens, when it comes to government. And nothing to do with temperament--difficult in the sense that there are millions of edge cases that have to be supported.
I read (and thoroughly enjoyed) Recoding America v recently, and throughout found myself thinking - this is basically an American, tech-focussed, and less Stafford Beer-y (consciously at least) version of Unaccountability Machine.