Tuesday, January 15, 2013

Ok, so, Bureaucracy

I was in the process of writing a post on how the government still doesn't know when I was born and the problems that that seems to be causing, when suddenly I got a call from that part of the government.  Which, I wonder if what happened is that the bureaucrat I talked with (who was on the national, not local, phonebank I believe), who let me know the government still doesn't know when I was born, responded with a WTF? and called up one of my local ones which in turn caused them to call me.  Because it seems very unlikely to be coincidence.

Now, this marks another thing against the bad bureaucrat from before (good god do I wish I'd gotten the good one two days in a row instead of being stuck with the bad one on the second day) because the last thing that happened before we parted was that she printed out a piece of paper with various information on it and asked me to verify it under penalty of similar-to-perjury and I had to say, no, it isn't correct, those two fours should be nines.  And I pointed out which two fours because other fours on the page really were supposed to be fours.

And then she asked me again to verify its accuracy under penalty of similar-to-perjury, "Is this information correct to the best of your knowledge," and I said, and I quote, "Except for the phone number."

Reason I haven't heard anything in over a month?  In spite of her claims of doing it right there and then on her computer, she never fixed the phone number.  So anyone trying to contact me ended up getting a law office or some such.

Then, apparently, it was guessed that this could be a case of 4-9 confusion and the call came in.  Or, if as I suspect calling up at the national level to try to understand resulted in movement at the local level, maybe whoever called the local office gave them the phone number I used to call the local level.

Anyway, the bureaucracy is too small, that's why you're left on hold and the waits are agonizing and no one ever likes the idea of having to go to the DMV (or in my state BMV).  There are not enough customer facing employees to handle the customer load.  Thus waiting.  But I wonder if the undersized bureaucracy has another effect too.  If there are only a few employees then getting rid of the ones who suck is getting rid of a fairly large part of your workforce.

If there's any gap between, "I've got to let you go," and hiring and training a replacement there are few enough people that either you're going to have to significantly increase the workload of those remaining or you're going to have to keep the person on which means you've got an employee who was crap to begin with but now is pissed off at you too.

If you've got twice as many employees you can cull the worst element from the herd while only increasing the load on those who remain until a replacement can be made by half as much.

The fewer employees you have, the more incentive you have to keep ones who are, objectively, crap but still remain largely adequate in their crappy way.  You don't have the scale to absorb their absence until you find and train a suitable replacement.

Anyway, I do support the bureaucracy in general because they do important and frustrating work, but there need to be more of them (wait times are atrocious) and they need to be better about getting the worst elements out.  The first is a much simpler thing to do than the second.  The first is served by just hiring more people (and train them of course.)  But the second...

If most people only deal with someone once or twice, how do you determine how well that someone is serving those people?  I don't know.



By the way, national phone bank person seemed as confused as I was as to why my original birth certificate, which I brought into the building, wasn't enough and the local felt it necessary to get a new one from the hospital directly.  Still have no answer as to that.  I wonder if they're allowed to make judgments on the spot of the form, "I don't like how this person looks, let's use as much scrutiny as we can."


  1. I don't think this is entirely valid, though. If there's a random distribution of skills among potential employees, then the amount of employees who are below skill level X will be a constant proportion of the workforce. If anything, it's harder to fire everyone who's below a certain skill cutoff, because that's more people and more paperwork.

    On one hand it's easier to fire the single least competent person from a larger pool, because they're a smaller proportion of the workload; on the other hand, chances are they're less competent than the least competent person in a smaller pool.

  2. I don't think this is entirely valid, though. If there's a random distribution of skills among potential employees, then the amount of employees who are below skill level X will be a constant proportion of the workforce.

    From what I've seen disposition is more important than skill level. If someone is below a certain skill level you can always give them more training, which still takes them away from doing their job, but not as much as having them stop doing that job, and it saves you the trouble of finding someone new.

    That said, assuming a random distribution there will always be a certain proportion of people in the potential workforce below a certain level, but that doesn't mean there will always be a certain proportion of people in your actual workforce below said level.

    Assume your goal is to have no one worse than average. Which is a kind of lofty goal (even though it doesn't sound like it) because half of all people are worse than average.

    If you cut your worst employee from a randomly distributed set then the odds are pretty good that just by picking at random you'll get someone better. If you have some kind of evaluation period where they're not fully hired yet just to make sure that they're not as bad or worse than the person they're replacing, then you can make sure you've improved your workforce.

    And the person you cut isn't doomed to forever be unemployed because their assets may better match another job, perhaps one that's even within your own company.

    But the question is can you do it. Consider customer facing employees. In fact, just consider those at the front desk.

    If you have one then you can't cut them until you already have a replacement, because if you do then there's no one to cover the desk.

    If you have two then you could theoretically cut them, but doing so would double the work of the person remaining.

    Three and cutting them means means 150% effort from the remaining ones until replacement is found.

    Four and it's 133%

    Five and it's 125%

    Six and it's 120%

    Seven and it's 116%

    Eight and it's 114%

    Nine and it's 112%

    So on, so forth. (And front desks can be huge, look at any large bank. Also, once you hit 4 people at the front desk there are things you can do to speed up the line beyond adding more people, Queueing theory is a branch of mathematics, they've made some interesting discoveries.)

    Anyway, assume you can afford to cut the worst person. If we assume that the hired population matches the available work force population then odds are that, selecting a person by random chance, the new hire will be better than the old one. Because the worst person comes from the crap tail of the curve, the majority of people waiting for the job are from the better section of the curve.

    And presumably you've got some screening process to make sure you don't hire people worse than your worst employee.

    [Math in next post due to character limits]

    1. But for the pure math of it:

      Assume a random distribution in the workforce. Assume hiring is done at random with no screening whatsoever so the people hired at any given time will match the distribution in the workforce. Assume that we're not considering the ease of hiring people, or the number of people currently employees. Assume the distribution in the potential workforce remains constant over time. (Simplifying assumptions, don't you love them?)

      Hire first group of employees. You have 50% below average, 50% above average.

      Set as the, "We have to let you go point," the current average. Do not change this point. (So that it will always remain the average of the potential workforce, which we have assumed to be constant.)

      First culling: lose all the people below average (not just the absolute worst as above), hire replacements at random.

      Of the replacements, because of the distribution of the potential workforce, half will be above average, half below.

      Your workforce now contains 75% above average workers, 25% below average.

      Next culling, let go the 25% below average, hire replacements at random. Your work force now contains 87.5% above average workers, 12.5 percent below average ones.

      Next culling, your workforce now contains 93.75% above average workers, 6.25% below average ones.

      Now, if everyone does this the average of the available workforce will end up going down, which will change the math, but also the new hires that are eventually culled get training, so they might be making the move from below to above average themselves, and it really does get terribly complicated, but the overall point is that there's no reason the distribution in your company needs to match the distribution in the potential workforce.

      If you do your work right it shouldn't unless we have at long last reached full employment, and even then maybe not because it could be that you find someone isn't good working the front desk, but is stellar at crunching numbers/shifting files/whatever. So the below average people aren't necessarily being dumped back into the potential workforce and, as noted, at least they're getting training and experience every time they do get hired, meaning that they may go from being below average to being above average over time causing the distribution in the potential workforce to not be lowered as much when they're put back in as it was raised by taking them out.

      Now if you're going for the top 2% then that's harder because 98% of potential employees don't meet your criteria, and you're going to need one hell of a screening process. (And money to compete because the top 2% is always going to be in demand.)