Yes, we can…stick our foot in it
By Daniel M. Ryan
Had he not been perceived as dangerous, President Obama's slogan "Yes, We Can!" might have been met with a knowing smile. Too many politicians have been down this trail, and it often ends in mud. Government is afflicted with something akin to the computer programmer's curse.
Put simply, the curse is: the more intuitive the program, the harder and longer it takes to code. Believe it or not, the hardest key to code in a simple text editor is the delete key – although the backspace key runs a close second. What makes the delete key so hard to program is the need to copy and move the text that's right after the character being deleted. Keeping the line order in line is another complication. In addition, the string-copying functions used to effect its operation are some of the commands most vulnerable to computer viruses. In order to safeguard the program, the copy operations have to be broken down character by character so as to decommissio n each byte of memory used for the operation – or new operations that do so have to be found and used. If those bytes aren't decommissioned, a hole is left for a virus to get in.
Less homely examples are legion. The grandest are natural-language processing and the related field of artificial intelligence. Coding work on both started long before the personal computer even existed, because they seemed like easy problems to solve. At first blush, natural-language processing procedures seemed little more difficult than calculation tasks. Expert systems seemed even easier than natural language, because the questions and responses are circumscribed to specific subjects.
And yet, from the '60s to now, general solutions for both are out of reach. Both are the inside straight of computer science.
Governmental measures that sound nice are subject to the same curse. How hard can it be to issue a check for something, if taxpayer funds are available? How hard can it be to offer a job to anyone who wants one? The mechanics of both are very simple – as simple as conjugating a common verb, or matching a word to its meaning. Divers into the subject can find divers objections to keeping it simple. The drug the objectors seem hooked to as they drug their feet seemed unshakable. Through their canned speeches, they almost invite being canned. It made those who hold elected office dream of chucking civil servants into the hold of a boat. How hard can it be?
As hard as, say, natural-language processing. Both are like bramble thickets, which get entangled through natural growth processes. Two incidents this week show clearly that Big Government often means the Big Bramble.
First: The AIG Cover-Up
Tim Geithner did not have a good week. Leaked documents from his time as head of the New York Fed show them urging – telling – AIG to redact material information from a draft SEC regulatory filing that disclosed paying 100 cents on the dollar to redeem $62.1 billion in credit-default swaps. Later clarification from the Treasury said that Geithner was not personally responsible for the redaction. A New York Times report paints a perplexing picture for those who think that business in inherently anti-regulation: "The e-mail exchange between the bailed-out insurance giant and its regulator portray a strange reversal of roles, with A.I.G. staff arguing for the disclosure of certain details on payments for credit-default swaps to major banks, only to be discouraged by officials at, or representing, the Federal Reserve."
Other defenses have been floated into the public sphere. The New York Fed has claimed that AIG was offered "advice" regarding the non-disclosure, even though the stroke of the pen through the item in question looks a lot like an order from a bailer-outer.
It's easy to see what went wrong. In the heat of the moment, the New York Fed didn't want the bailout responsibilities to multiply. Since the banking system of today rests on confidence, the best way to assure a relatively smooth outcome seemed to be suppressing any items that would impugn market confidence in either AIG or the bailout. A simple redaction…
…that flew right in the face of long-standing SEC disclosure laws. This scandal shows that one branch of the bramble bush did get tangled with another. It's quite clear that the "simple" bailout is a lot like "simple" natural-language processing.
Second: Maximum Feasible Overreach
This one has been far less costly but far more terrifying. Soon after the Christmas Day plot by Umar Abdulmultallab was foiled by heroic passengers on Flight 253, it was revealed that Abdulmultallab's own father had warned the U.S. government about his son's turn to radicalism. Later investigations revealed that intelligence sub-departments got warning signs, including that one, but didn't share with the others and connect the dots. Consequently, what was a serious threat appeared as another routine false alarm to each sub-department.
It's the bramble bush again, as this Yahoo! News report makes clear:
President Barack Obama "has now discovered that he's inherited an intelligence community in the United States which is bloated, bureaucratic and even with the best of intentions has become so large it finds it very hard to put together the pieces," Bruce Riedel, a former CIA officer, told AFP….
"There are two dozen people in that room. Why are that many people trying to run the show? Why doesn't he have an intelligence senior official who he goes to who is in charge of everything?" he said.
Clearly, the U.S. intelligence service is a victim of specialization. Unfortunately for Adam-Smith-lovers, intelligence is not like a pin factory. The tasks in the former don't overlap, whereas the more complex tasks in intelligence do overlap. We've seen so.
Unfortunately, Reidel's proffered solution opens up a different can of worms. If the information-sharing is not to balloon into information overload, there has to be limits on what's shared. There's a risk that future warning signs will stay stuck in the sieve.
Third: The Next One Coming
There was a lot of cheering over the health-care-reform amendment co-sponsored by Senators Jay Rockefeller and Al Franken. It limited non-medical-care payments by insurance companies to 20% of total payments. Everything else – not just profit (if any), but administrative and corporate costs – have to be under 20% of total disbursements. A simple task, right? Surely, there aren't divers objectives therein.
In fact, the idea is eminently transferable. I humbly suggest, as a trial balloon, proclaiming that the exact same restriction be applied to welfare services. 80% of all disbursements to welfare agencies should go to welfare recipients. All other disbursements should be capped at 20% of funding. Since welfare agencies are non-profits, profit won't even enter the equations; the cap would only apply to administrative and head-office costs. I'd like to repeat that I'm not suggesting enacting it; I'm suggesting proposing it, along with an invite for welfare-services professionals to register objections to the idea. Later, when the insurance-company cap is in force, those objections will prove to be uncannily prophetic.
Afterword: Knowledge And Common Sense
It's sometimes claimed that the bramble bush problem can be solved by use of common sense. People should just use their heads a little better. Unfortunately, common sense has been elbowed out to a larger degree than we know. Look at how many theorems in political economics – moral hazard, the free rider problem, the tragedy of the commons, etc. – are merely codifications of well-worn common sense. If common sense has had to be reconstructed along academic lines, then some was lost along the way. It would be fruitful to ask why, because we might be losing some needed common sense today.
Daniel M. Ryan is currently watching The Gold Bubble.
Get weekly updates about new issues of ESR!