Tuesday, August 30, 2011

Overlearning the Game

A good article on politics and corruption:

Short on proposed solutions. Here are two:

1. The Severe/Ben Franklin approach: The founding fathers were in a time of revolution, and many were quoted as expecting more revolutions to occur that would throw out our Constitution in favor of an even better one in 50 years or so. With this you also throw out all the corruption of those who have not only "overlearned the game," but also those who have invested massive amounts of wealth to change society's rules so its wealth is diverted back to them. This seemed likely to those in the midst of a revolution, but not very likely to us, and, well... what would happen to all my stuff?

2. The Gradual/Videogame Cheater approach: Game authors attempting to fight cheating fight on a naive technical front first, and later a complex behavioral front second. The technical front is like many of the basics of campaign finance laws, like "You can't pay people to vote for you." The obvious hacks. But you eventually realize you're not really trying to stop obvious hacks, you're trying to stop creative, insidious ones too. So you model good behavior, and basically treat everything outside of it as cheating. The challenge is to include modeling very strong performance in that model so you don't risk punishing it if it arises legitimately.

Google is arguably working against SEO entities who "overlearn the game" and either google bomb or push their company's result to the top. The big difference between Google/Videogame companies and law is how nimble they are. The companies acknowledge there will be cheaters and have teams to quickly respond to new workarounds for the system. The law sits idle and a big messy Congress that can be corrupted by cheaters themselves is in charge of fixing it, in a really ugly process. Arguably Congress should continue to set the broad strokes like "Prevent campaign finance abuse," but a nimble more company-like organization within the government ought to be responsible for implementing that and quickly responding to new abuses.

Wall Street actually has a limited version of this model - the SEC is not an elected body - but it's somewhat limited in scope, and impotent when for example its powers are delegated to the OMB when it comes to Collateral Debt Obligations, and down comes the world economy.

I think Congress might be suited to setting the powers more broadly and the metrics for success on these nimble organizations, but they definitely aren't suited to creating the laws that actually trap the cheaters. They're just too slow, by design.

Thursday, August 11, 2011

No Student Loans for Low-Pay Degrees

In 2015, if a private college can't show that at least 35% of graduates with a given degree can pay back their loans, students entering that major won't be eligible for student loans. Pretty smart program.

Not surprisingly, private colleges are upset about it - they want that free money to keep people entering majors they know don't lead many kids to successful careers.

It would be nice if a law went farther, and required students entering a major or considering majors to be informed of:
  • Average employment rate
  • Average employment rate within the chosen field
  • Average salary in 1 year and 5 years
  • Average time to repay
  • Average delinquency rate
of past graduates as a whole. It's fine if people want to say they went to college for intangible hard-to-measure reasons, but when it's putting you in a tangible $100,000+ in debt, it's time to be serious.

Tuesday, August 9, 2011

Dying soon?

Dying soon? Put your life story into your grave, so people can listen to it with their cellphones when they visit:
http://www.npr.org/2011/05/30/136676964/technology-brings-digital-memories-to-grave-sites

MIT actually predicted this 40 years ago with Project Oxygen: http://oxygen.lcs.mit.edu/Overview.html

We would someday use "handheld devices" to get "location-aware data" about our environment, like whether there's a meeting going on on the other side of a closed door. In their version though, the data about the dead person was stored in the grave itself, rather than just having that position, RFID tag, or QR code associated with a website about the corpse.