Thread by @lymanstoneky, There is a relationship between these, however. Though it's often extremely complicated. [...]

There is a relationship between these, however. Though it's often extremely complicated. The issue is to get competent bureaucrats you often need fewer of them, so that they can actually have agency rather than be mired in committee. https://twitter.com/Noahpinion/status/1335267609621659648

https://twitter.com/Noahpinion/status/1335267609621659648

One time I found an error in a program. It was not a big error (I found some big errors a different time, not this time). I wanted to fix the error. Had it been me the analyst and another person the administrator we could have fixed it in 5 minutes.

We'd have noted the error, made the fix, sent it to the lawyers to write it up, and we'd call it a good day with a few million dollars of taxpayer money saved.

But that is not how it worked. Because my analytic assessment had to be reviewed by people who were in the review positions because they got to a certain GS grade in another area and transferred laterally.... so did not understand the basically quantitative issue.

And it's not just that. Once I found the error, everybody with similar programs had to go look for it. That makes total sense, of course; but it meant we couldn't fix the error I found until we knew if they had it too: so that we could all be consistent.

That totally makes sense from a very literally bureaucratic perspective. You don't want to have a two-year-long sequence of correlated program changes.

But, wait.... why not?

The reason you have to make all the changes at once is because it makes you look competent, whereas finding one which leads to external stakeholders finding related errors makes you look incompetent. Specifcally, makes higher-ups look that way.

Stakeholders complain, etc.

But the perception here is clearly backwards! Delaying the fix of a problem because you want to first do a CYA on potentially similar problems is not "competence"! It's spending millions of dollars unnecessarily!

And in principle, making a fix and then publicly saying, "This problem might impact other programs, prepare accordingly and maybe do your own homework and see if it's gonna impact you" is a very efficient thing to say to stakeholders. It offloads some work onto them!

It also means the change is smaller at T(o), and can help manage expectations at T(n). Arguably, this is less disruptive than changing five or six programs all at once.

The end result of my story is even more irritating. The error did not impact any other programs. Basically, a legacy spreadsheet problem related to the Lotus-Excel changeover like 15 years ago in a program I was indirectly related to had led to unjustified payments.

I flagged this. After a few months of analysts checking other programs, there was no error. Which naw duh because it was a legacy spreadsheet problem and they had literally different files and file formats.

But it took months because it was a tasker on each analyst's to-do list and it had differential priority for each (this is unavoidable) and so the next step couldn't proceed until each analyst had gotten around to it and concluded it was not a problem for them.

The issue here is NOT bad analysts. Nor is it even bad managers. Everybody did everything right. Nobody made a mistake in the process. It's just that the process was not a good one.

Also a lot of the paperwork had to be hand couriered around the building for triplicate signatures which was just wild.

Anyways after we were all analytically on board with "guys, it was a spreadsheet error, this is not that complicated," we then had to go to the actual decisionmaking level with our nice little memo saying "Lotus sucks!" and get them to actually ;et the policy change

The issue is the decisionmakers often are woefully unprepared for the complexity of modern government. And this isn't even political appointees; this is professional staff! But professional staff with long and distinguished basically nontechnical careers.

And often professional staff who are very good at some things, maybe they were awesome at managing some program a decade ago, end up in positions they just aren't great at.

And so you get my humble "Seriouslly, Lotus is bad!" memo in front of a decisionmaker who would now like to "hear the counterargument" for why we should in fact keep the spreadsheet error.

Now this is a good case of why we should NOT have done all the analysis in advance. Because the decisionmaker was always going to demand that we Have The Debate. So by resolving the debate in advance we shot ourselves in the foot.

So anyways then we have to come up with an argument for why spreadsheet errors are good, actually, and maybe the error was actually unintentionally the Ultimate True Intended Function of the program.

And since the decisionmaker is very busy, even if it only takes me a day to bang out a memo on the counterargument and then address that, it takes another month and a half or so to get another meeting.

So we finally get the meeting. And it's literally like 2 minutes. Change approved. We send it to the lawyers to draft and it's done in like a day. Then it's the regulatory review process and out of my hands and that process is what it is.

But the point is that we could have noticed the error and immediately started the public process, doing our internal process simultaneous with the public one. More to the point, we could have streamlined our process.

But we could not actually do this because if we did this we'd have to fire a lot of managers and push decision rights downwards. Because we have managers who need to justify their existence.

What you really need is:
1) More provisional/first line decision rights located with technical experts
2) More retention of technical experts without changing their jobs

The problems with this are:
1) A lot of government programs do not require any technical expertise but bureaucrats claim to be "experts" when in fact their job does not require much of any expertise

2) Political oversight is necessarily top-down

Dealing with 1) requires tackling the GS system, unions, and public perception. The reality is a lot of government jobs do not actually demand any technical qualification. Patronage is bad, but listing qualifications isn't getting you more competent bureaucrats.

In some cases, meritocracy can produce worse bureaucrats, because you get people who basically just need to read over some applications and compare them to some diagnostic criteria and one or two qualitative tests, but who conceive of themselves as experts.

I'm not saying we should bring back patronage. But I am saying automating away a bunch of these jobs would be a huge benefit to society.

For 2) there's just no easy solution. Political oversight is what it is. Political decisionmakers want to understand their decision and especially how it will be received. We live in a republic where those decisionmakers are the representatives of the people and thus in charge.

Which is to say, government bureaucracies are just gonna have problems that can't be fully resolved.

But some problems can be addressed. The incompetence of the vital statistics bureaucracies in the US has been clear for a long time to anybody who works closely with that data. COVID has put it on display. CDC's whole bureaucracy crapped out.

Market failure exists. So does government failure. And a lot of government failure arises from smart people in dumb processes that push too many decisions too far up the food chain.

Latest Threads Unrolled: