07 November 2010

Weird conclusions

I've been reading recently about abductive reasoning and on-line learning in a formal sense, and as an ancillary I've been thinking a bit about how people update hypotheses, discarding or modifying them as they gather new data.  Mostly I'm just struck by how terrible many people are at this.

Two recent news stories in particular made me think of this.

The first are those explosives that were found in air freight last week.  I lost track of how many people used this as evidence that we need more TSA screening of passengers.*  They did not seem to realize or care that this new attack was entirely disjoint from passenger screening.  It's a bit like discovering a burglar had attempted to sneak in through a window and deciding that you should go out and put stronger locks on the doors.

Some people at least used this as a reason to advocate more screening of air freight.  At least they're in the right ballpark, but these devices weren't discovered because of screening.  They were discovered by Good Old Fashioned Police Work.  In fact almost all terrorist plots that have been averted have been done so because of investigation, but people rarely advocate more investigation, and instead agitate for whirligig technologies that didn't have anything to do with the incident.

Some people doing this must recognize the inconsistency and are ignoring it in order to further their political ends, but many people seem to have a legitimate cognitive failing updating their beliefs.

The second story is this one in the Economist about non-ethanol biofuels.  Executive summary: without cheap sugar cane, ethanol as a fuel is a foolish proposition, making it silly for everyone except Brazil, more or less.  And yet America (and many other countries) throw money at their corn farmers to make wasteful ethanol.  Future technologies with alternative feedstocks and making heavy use of genetically modified bacteria and algae to make "drop-in hydrocarbons" are very promising though.

(BTW that's a really fascinating article.  I recommend it strongly.  Genetic engineering is the future of fuel.)

Some people have used the tech outlined in this article as evidence for the wasteful corn-ethanol program we have now.*  I've seen this with reports about a lot of other reports about how alternative energy isn't efficient now, but could be in the future -- people use this to conclude we should invest billions in building out today's tech.

Actually the very same intelligent, educated chemist I mentioned in this post told me that we should subsidize residential solar panels now even though they aren't cost effective specifically because innovation should lead to more efficient and cost-effective solar panels in the future.  I said that if he was confident that future solar cells will be effective while today's aren't you should actually be paying people to delay installing the less-effective units of today.

I think people just grasp onto these hypotheses like "we need more screening for terrorists" and "we need more alternative energy" and they take any new data that is vaguely related to this in a positive way and decide to strengthen their confidence in their original hypothesis whether or not it actually does.  I think this may be because people don't have a specific mental frame about what and how these things work, they just know that they're the sorts of people that want more anti-terrorist security or more alternative energy.  It's more about their identity than any specific idea.  So instead of needing a new data point about screening preventing terrorism, for instance, they're satisfied with a data point that says "terrorism exists."  Similarly they don't wait for data that says the specific energy programs being subsidized are beneficial, they just need some energy program somewhere to be worthwhile.  They don't bother conceived of individual programs or theories, they just dump everything into one big bucket.


* I had some links to examples of both these silly reactions but lost them, and I don't feel like wading into the stupid to find them again.  So just trust me that these are common sentiments if you please.


PS I see this same thinking in drug and alcohol policy a lot.

No comments:

Post a Comment