Monday, March 14, 2011

Congress Has Never Used Shark

Shark is a performance profiling tool that I use heavily at work. In a nutshell: Shark tells you what part of your program takes the most time to run, so that when you spend time making your program faster, you make the parts that matter run faster.

To construct a trivial example, imagine that my recipe program spends 99% of its time drawing beautiful 3-d recipe cards, and 1% of its time drawing the menu bar. If I want it to be faster, I have to make the cards faster, not the menu bar. If I make the cards twice as fast, the program is twice as fast. But...if I make the menu bar twice as fast, you'll never notice, because I've only affected 1% of the total problem.

Put another way: it pays to go after the whale, not the minnows.

Consistently going after whales is what has made X-Plane fast over the years.

So it drives me nuts when Congress goes after minnows.

Here are a few lists of potential spending cuts...we'll see how much of this actually happens. There are even some cuts that save more than $1 billion - that is, cuts that almost matter.
  • $1.1B - Office of Science.
  • $1.7B - GSA Federal Buildings Fund
  • $1.6B - EPA
  • $1.4B - DOE Loan Guarnatee Authority
  • $2B - Job Training Programs
  • $1.3B - Community Health Centers
  • $1B - NIH
  • $1B - High Speed Rail
(And we have some minnows: $6m for the NEA, $6m for the NEH, $7.3m for the Smithsonian, $2.3M for Juvenile Justice...if ever we optimized the menubar.)

But the real problem here is the total failure to 'Shark'. A few big ticket items seem to have strangely escaped the list.
I'm pretty okay with the 'less government' idea...but I'm not okay with the 'less government except for the stuff I like' idea, and either way if you're going cut, make cuts that matter.

(Why is Social Security not on this list? Social Security takes in a lot of payroll tax, and could be made solvent by relatively small changes in benefits, the payroll tax cutoff, and retirement age. Heck, Social Security was doing fine before Wall Street cratered the economy. Compare this to Medicare, which has jumped 13% in cost in two years.)


Jonathan said...

Not every program's budget impact is best expressed in terms of its yearly spending. Some programs have a much bigger impact than they look like they have. For instance, that loan guarantee fund figure under-represents the risks if the loans all go belly-up. I'd be happy to see that program get axed.

Rissa's uncle works for DoD, trying to remove wasteful spending. The amount of stuff he finds, and the difficulty he has cleaning it up, tells me that DoD spending is a biiiiig target and Congress should really be examining it more closely.

Benjamin Supnik said...

re: DOD - you don't have to tell us...Lori and I lived in Maryland for five years, I would describe the "efficiency" of DoD funding as an open secret.

Re: guarantees, totally true...that's why Geitner can walk around with a straight face and say "the bailout was really cheap" and conservative pundits will claim "fanny and freddy were the most expensive part of the bailout" - because a cash-only accounting doesn't take into account risk premiums or their real market value.

What you'd really have to do is apply GAAP and treat every guarantee as a liability with a fair market value. In the case of the bond market, some economists have applied this metric, by looking at (1) how much government backing lowered borrowing costs for sick banks and (2) how much it would cost to simply buy that much credit protection via the OTC market. The short answer is: it's a huge subsidy and it has a real cost.