Confession of an IQ Hacker

Intro

I’ve never taken an IQ test, at least not unofficially, that is, of the HRIQ type test. Now, before you get your hair all up on your back thinking I’ve cheated on the some 40+ odd unofficial IQ tests taken, think again. With that said, maybe before I start, I should give a minor bio, at least as it relates to this essay.

Bio

My name is Kenneth Myers. I was two years old in 64, 1964. When I was small, all I liked to do was tinker or “hack” (as in the MIT definition, not the cybernasty kind) at things. That is, I was someone with playful technical ingenuity and spirit. Too, I detested academics, or anything for that matter, taught by another. In other words, I enjoyed the art of self discovery and experimentation. Each of these philosophical “moods” spilled over into every aspect of my life. From hacking at TV sets, to hacking at parents. You name it, I hacked at it.

In my life I’ve taken exactly four official IQ tests, two of them noted, two of them not. The first test was the SBLM at six years of age. On this, my stepfather later informed me, I scored 181, ratio IQ. So, this doesn’t really count, although, it was officially annotated. The second test I took was the WAIS as a young adult. My girlfriend at the time was studying for her Master’s in psychology and asked if I could be a guinea pig. I agreed. I scored a 140 full scale IQ, with a verbal IQ of 125 and to her, my girlfriend’s that is, amazement, a 152 working memory IQ. This test was not documented, but simply for her Master’s thesis. Shortly after this time, my girlfriend and I went our separate ways, I joining the military, her becoming a therapist. Just before my separation from the military, I was again given an IQ test, the MAB it was called; and we, that is, the group chosen for taking this test, were the norming sample. This test, I scored 148 full scale IQ. This too, was not documented in our military transcripts; as it was just a norming sample. The last official IQ test I took, was the CTMM. That was for entrance into Mensa. I scored a standard deviation 15 equivalent of 131. However, this test was capped at 150+, Mensa scoring not including the extended norms for this test. I believe I aced the nonlanguage portion of this test. That’s it for official tests.

I eventually did some academics; graduating with a double major in physics and mathematics and attending graduate school for a master’s in mathematical physics. I later worked in the computer industry, from software engineering to computational science.

Hacking IQ Tests

My first HRIQ test was the Mega test from an old issue of Omni magazine. I decided to take the test. I did fine, scoring the equivalent of my MAB score noted above. However, this is where I noticed something different. Unlike the tests listed above, I could use any resource I pleased in taking this test; any resource. This any, in and of itself became quite intriguing to me. About this time, I recall purchasing an electronics magazine that contained an article on neural networks. I was a changed person. Here was, for the first time in my life, apart from watching the movies Colossus the Forbin Project and 2001 A Space Odyssey featuring, HAL (IBM) 9000 as a kid, artificial intelligence writ large. Interesting, to say the least. Then, to my amazement, I came across John Searle’s book, MINDS, BRAINS AND SCIENCE.

This is where my attitude, mood, changed towards HRIQ tests and IQ testing in general; this is where I discovered the, Chinese room argument. Here’s the gist; the thought experiment...

Imagine a person who knows no Chinese locked in a room with a massive book of English instructions. No more, no less.

The Process: People outside slide slips of paper with Chinese characters under the door. The person inside uses the English manual to look up which Chinese characters to write in response and slides them back out.

The Result: To those outside, it appears the person in the room understands Chinese perfectly.

Conclusion: The person in the room still doesn’t understand a word of Chinese—they are just following a program. Searle argues that digital computers are in the exact same position: they process syntax (the form of symbols) but have no grasp of semantics (the meaning of symbols).

Searle concludes that computers simulate rather than understand mentalese. However, this was not the big deal to me. What intrigued me were the process and the result, namely, test makers (people) slide tests into magazines, periodicals, or the internet. A person uses any, again, any resources (except other people) they please to complete the test. The results; to the test maker it appears the person who took the test; understood the test. My hack. I wanted, as I’d partially realized when taking the Mega test and using a university library as my resource, to develop this concept further and, in essence, automate my test taking. I wanted, in essence, to test the IQ of the library, not my IQ.

Over time and not possessing much computer access, I developed methods for taking these tests; involving everything from correlation and cross referencing of material. For instance, given an analogy’s test or section of a test, I would;

1. Look up the individual terms of the analogies.
2. Cross reference these within a crossword puzzle or scrabble dictionary available in the library.
3. Rank order the terms that overlapped between definitions.
4. Generally, one to two terms would fall out of the algorithm.
5. If one, this was the answer.
6. If more than one result, goto 1.
7. After a few recursions, this method of successive approximations would generally converge, especially on these first generation HRIQ tests. It’s similar to looking up a word in the dictionary. Start with the first letter, land on P for instance. If the word is PETAL, go to PE. Whoops. PF, go back...etc.

Over the years I’ve perfected these methods, using any and all resources available, developing them if not. Most were available. Like today’s LLM for instance.

You see. I already knew my IQ from the four tests I’d mentioned in the BIO. So, what did I care about my score? I just wanted to see how far I could take the Chinese room argument and get IQ scores for libraries, supervised and unsupervised algorithms, numerical methods, Sloane’s integer sequences, and the list goes on. After perfecting the application of many of the methods, the operation became almost turnkey, that is, all necessary and ready for immediate use.

Using these methods, and disregarding LLMs, I’ve managed IQ scores ranging from 125 to 170, by deviation 15; and all basically, with the push of a button. More recently, around 2025, I integrated LLMs into my model. With this addition, I’ve seen HRIQ tests jump in complexity; albeit, horizontal complexity, not vertical.

Let me explain. According to the model of hierarchical complexity; “intelligence” is simply the ability to complete tasks of a certain hierarchical complexity. These tasks are rank ordered “vertically” according to their complexity. For a task to be more complex than a previous one, it must meet three criteria:

1. It is defined in terms of two or more simpler, lower-level tasks.
2. It organizes those simpler tasks.
3. The organization is non-arbitrary.

The MHC defines roughly 14 or fifteen of these orders. Look MHC up if interested. But for us; the MHC suggests that “growth” is the process of learning how to take these lower-level actions and organize them into increasingly sophisticated, higher-level “hacks” or systems. In my case, these hacks became a salvo, and as above, eventually a turnkey operation that worked for years.

However, these days much of the additional complexity and difficulty of tasks in HRIQ testing comes from either knowledge of rare items, oddly coupled sequences, test take, retake practice, psychological profiling and honestly, just pure noise. While HRIQ tests have become a popular and common pastime measurement of how “smart” an already superintelligent person is, whether they are the best measure of this construct is up for debate. Much debate in fact!

But. One thing I will say, is; given the advent of LLMs my days of “hacking” HRIQ tests are over...because, honestly, LLMs are the ultimate turnkey operation. And, they are here to stay.

Kenneth Myers