The Intelligence Bubble: A New IT Bubble, or More?

• John Vandivier

This article argues that there may be a bubble in intelligence, and it also clarifies why there may not be a bubble in intelligence. We also define what we mean by intelligence.

I am using the general term intelligence to refer to three things:

  1. Business intelligence including big data, analytics, and so on.
  2. Public intelligence including the CIA, NSA, DIA, and so on.
  3. The notion of individual intelligence, education, and the education system.
I previously argued that there is a small bubble in the wages of data scientists which may either be attributed to low supply or high demand and it perhaps reflects a general bubble in business intelligence.

Do I really need to argue that public intelligence is over rated? I will just state it as a fact and leave the debate to others:

  1. The NSA doesn't stop terrorism.
  2. Bad intel lead to the war in Iraq.
  3. The CIA has been corrupt since it began.
  4. Central government in general provides for the moral hazard of special interests due to concentrated benefits with disbursed or diffused costs. This can be called the problem of special privilege or political capture.
  5. Central government in general provides for inefficient (read: not intelligent) selection due to the calculation problem. This problem applies to any society with reduced access to optimal information, not just Communist countries. Any society which has any sort of market distortion will suffer a calculation problem based on price information. In addition, any decision making structure which has fewer minds than a full market will have a reduced number of information processors as well as less access to the optimal amount of most accurate information. This means that even a country as open as the U.S. suffers a 3-fold calculation problem. The only solution is a polycentric and competitive market for government.
Next I'll argue that there is also an education bubble and that personal intelligence is a myth:
  1. Traditional education seems to predominantly be a signaling game according to Bryan Caplan. TIL an old school term for signaling is screening. Check out section 5 of this paper from 1994.
  2. Google and others note that college degrees have no statistical significance as a predictor of performance after 3 years out of college. These companies prefer experience and demonstrated skill as evidenced by something like a portfolio. Are companies increasingly accepting badges and skills-specific credentials? I hope so.
  3. Grades are inflating. They are inflated even after accounting for learning efficiency gains.
  4. Public student achievement has remained essentially unchanged even as billions more are spent.
  5. Home schooling has been shown as the most efficient way to boost student achievement, but it continues to be ignored, accumulating massive social opportunity cost.
  6. As with the public school story, state universities continue to run up costs while grades inflate and achievement runs flat. The price inflation is fed by state subsidy.
  7. On the bright side, we have seen an increase in low price for-profit education options and alternative credentials like badges or certifications.
So public education is definitely a bubble but higher education is mixed and there is some fundamental improvement in some areas. Education per se is not in a bubble, but certain forms are.

Last is the notion of personal intelligence. I do think it exists but I think it's overrated for a number of reasons:

  1. According to leading effort psychologist Duckworth, grit and self-control better explain success than measures of talent. Grit is also a better predictor of success than IQ. Check this Ted talk as well.
  2. As mentioned in this section 4.6, IQ scores correlate with academic achievement but it's not clearly causal.
  3. The hypothesized g factor does seem to exist, but typically explains only about 40-50% of the between-individual variance in IQ test performance. IQ also varies with age.
  4. I think the g factor is largely explained by Duckworth's grit and self-control. In other words, I think intelligence is in large part a set of trained skills, and a product of grit and self-control. This trained intelligence would be a set of mental skills which are so highly transferable that they seem to be perfectly transferable, or a general intelligence. If true, the hypothesis that most intelligence is a set of trainable skills would unify Gardner's theory of multiple intelligences with other g factor research.
  5. While I haven't seen much evidence, on reflection I do think some portion of intelligence may be explained as genetic, in a way akin to physical genetic differences, but the size of that portion seems to be dwarfed by, grit, self-control, and trained skills including trained intelligence, if it is there at all.
In conclusion, there might be a general overvaluation of intelligence in our society.