Hacker News Reader

Comment by briholt | Hacker News Reader

Another major glaring problem that no one wants to admit is that lots of tech work requires (at the very least) college-level-IQ, which excludes the majority of the population. And the lower-IQ jobs are quickly being automated away.

View on HN

The vast, vast majority of tech work doesn't require any exceptional intelligence. It just requires a system of thinking, one which can be...trained.

... and the ability to be trained into learning arbitrary skills is IQ. Higher IQ means its easier to train yourself into being able to use a new system of thinking.

Like, being a soldier doesn't require any sort of exceptional intelligence either. It doesn't even require average intelligence. What it requires is an IQ of 85 - below that point, and the US army cannot effectively train you to become a soldier. Despite having every incentive in the world, the US army still rejects about 15% of the population.

15% of the population doesn't apply to join the Army. And those that do self select. Using the Army rejection rates for determining the IQ level of the population isn't a good method.

It's the definition of IQ. 15 points is 1 standard deviation, so about 15% of the population has an IQ of 85 or below by definition.

Just following up to add that these two 15's are unrelated, in case any readers get the wrong idea.

The area under the normal distribution curve from minus infinity to -1 (one standard deviation left of centre) is about 0.1586 or about 15-16%.

The 15 points = 1 stddev property of IQ is arbitrarily established.

IQ just means potential not capability. Given enought time I believe most people can be trained to do almost any job.

I don't think our brain suddenly evolved in 200 years from illiterate peasants to software engineers. We just have more school time these days.

Since it is empirically obvious that 200 years ago the world was not entirely populated by illiterate peasants ...perhaps your population model of the distribution of IQ and causes thereof is inaccurate.

Not entirely but the vast majority. It's empirically obvious writting existed 200 years ago therefore someone somewhere knew how to write.

What do you think people did before the industrial revolution?

A lot.

> IQ just means potential not capability.

Ok, you can say that.

> Given enough[] time I believe most people can be trained to do almost any job.

That's a statement about potential, not capability.

It probably helps but saying it's necessary seems a little far fetched to me. Also since pretty much everyone I knew growing up went to college I don't think "college-level" is really meaningful for much except wealth nowadays

That may say as much about you as it does about the population at large, right now about a third of adults in the US have a bachelor's degree[1].

1: http://thehill.com/homenews/state-watch/326995-census-more-a...

I would guess the vast majority of people in my highschool were in the top fourth of income in America (including my family, we were well off but not super rich because my dad was in the military and then a commercial pilot and my mom was a teacher). I'm sure other people have different experiences but waaaayyyy more than the national average of people from my highschool went to college and I can tell you that while it was a decent school it was not filled with geniuses. It's why I think a large amount of the reason people go to college is cultural and that to many people college ends up being more of a symbol of status than a intellectual achievement.

That's not true. It's a common complaint among engineers that developing line-of-business or CRUD applications is tedious and not challenging enough as it doesn't require a lot of creative thinking.

Truth of the matter is that most tech work doesn't involve AI, machine learning, self-driving cars or augmented reality but down-to-earth business applications. Developing those requires abstract thinking, empathy and problem-solving skills but it doesn't necessarily require a college-level IQ.

In fact a high IQ could even be harmful in that situation because apart from getting bored quickly highly intelligent people can display a tendency to overthink problems (which is probably how many notorious enterprise frameworks came about ...).

You've clearly never met a person who is unable to place a hanger inside of a shirt and hang it on a rack, or a person who can't sort 5 single-digit numbers mentally. What you are describing are 120+ IQ problems.

You make this sound like someone with an IQ less than 120 is mentally impaired.

An IQ of 100 is defined as the median IQ level for a population. The range between 90 and 130 covers the whole gamut of human intelligence that's commonly considered normal.

Much of the day-to-day work in IT often doesn't need original thinking but merely skillful application of known methods and patterns, which in turn doesn't require a college education.

Find one successful programmer at Google/Apple/Facebook/Cisco/etc. with an IQ of 90. Remembering basic equations, keeping long sequences of functions sorted in your head, even understanding FTP/Git/CL instructions require this level cognitive ability. When you deal with people who can't hang shirts you'll start to understand this. There are just under 200 million Americans with IQs between "can't hang shirts" and "productive programmer."

> Find one successful programmer at Google/Apple/Facebook/Cisco/

Point proven, eh? I'm not talking about these people (though I surmise there are quite a few in these companies as well who don't have that level of intelligence and simply tag along ...) but about the vast majority of IT workers working on some supposedly 'boring' database application.

Google and Apple in particular have been known for hiring highly qualified engineers only to then have them maintain some run-of-the-mill administrative software they're vastly overqualified to work on.

>doesn't require a lot of creative thinking

It doesn't, but you still need "higher level" developers to cover security and concurrent/parallel/distributed problems and maybe architecture.

It's not that someone can't also learn that stuff, but exposing a product publicly without some experience in those areas is asking for trouble.

IMO, Software is resisting division of labor and work by collapsing roles into "Full stack devs" or "DevOps Engineers".

I agree there have to be different skill levels but I think that distinction has to occur within denominations like 'full stack' and 'DevOps'. These labels simply determine what you do not how skilled you're at it.

In my opinion, full stack and DevOps is the normal, sane way of approaching software development. Artificially dividing up roles into labels such as 'front-end', 'back-end', 'database programmer', 'system administrator' only leads to more silos, less collaboration and sometimes even downright hostility between these roles.

IQ is standardized to a population distribution. As such someone with an IQ of 120 is not 50% smarter than someone with an IQ of 80; rather, they simply have a 50% higher "score". But there is no standard zero-point on the IQ scale; the "IQ of a rock" could be anywhere from -5000 to +50 or so. That doesn't mean IQ is meaningless -- it correlates with a variety of positive outcomes -- but that low-IQ people are not "proportionally" less intelligent as measured by the IQ. Rather the IQ scale is set up to have a seemingly normal distribution relative to the observed distribution of fluid reasoning in the human population. The standard deviation of IQs is arbitrarily set at 15, and the mean is set at 100. We could easily reform the IQ scale to have a mean of 1000 and a sigma of 1, or a mean of zero and a sigma of 100. The number, itself, is meaningless without context.

There is no evidence that a person with low IQ cannot accomplish the the tasks involved in e.g. software engineering, although they would probably do so more slowly than a person with high IQ. From Wiki:

"The prevailing view among academics is that it is largely through the quicker acquisition of job-relevant knowledge that higher IQ mediates job performance. "

Also, going to college has no effect on IQ, so "college-level IQ" is meaningless gobbledygook.

Re. college level IQ, I think they were referring to the level of IQ required to be able to successfully complete a college degree, ie. keep up with courses (which requires quick acquisition of knowledge), rather than the level of IQ 'acquired' after completing a college degree.

What exactly is a college-level IQ? Does everybody who goes to college have a college-level IQ?

A college-level IQ these days means something low like 100 -- you're probably thinking of a more stringent standard.

An IQ of 100 is supposed to represent the mean score of the general population.

The average IQ of someone with a college degree is a good bit higher than that. From a quick Google search it looks like it's around 115, which is a standard deviation higher than the general population.

I was referring to more of a minimum than an average.

I was under the impression among people with undergraduate degrees it was closer to 110-115.

Isn't half the population below 100? How can you consider that low?

That's lower than half the population.

Because it's terrible.

Pretty much the same way we can consider the median personal income ($31,100 in 2016 - https://fred.stlouisfed.org/series/MEPAINUSA646N ) "low". Because it's low.

The vast majority of the population already has a "college-level-iq" for whatever definition of. Tech work is not a geniuses' only game.

Maybe - or maybe it just requires a specific way of thinking about things i.e. Problem Solving and Time/Task Management? As well as the ability to learn the task of course.

What's the difference between "problem solving" and "IQ" ? They seem like almost the same thing.

"IQ" is a scalar semi-objectice well-defined metric that predicts some things about the much more complicated and less-well-defined concept of problem solving.

Problem-solving: the ability to deconstruct a problem into atomic pieces and construct solutions

IQ: How fast you do it.

tl;dr: Problem-solving is software, IQ is hardware

"IQ" measures something called "G", which is pattern-recognition ability (coincidentally the cutting edge of today's ML neural nets). The scientific theory of IQ is that that if you are good at pattern recognition, then you will be good at a wide range of cognitive tasks.

What is it about tech work that requires such a high intelligence, as you state? Because I'm a tech employee and would argue that your claim is completely unfounded. There's nothing special about learning how to code—it's a very trainable skill. Takes a lot of time and dedication, sure, but trainable nonetheless.

You can't even teach calculus to some people to the point where they can pass a class. Some people in programming classes can't even submit homework that compiles.

Calculus and programming are in two different leagues of complexity, and most programming doesn't require a knowledge of calculus.

Like any other skill, programming comes more easily to some than others, but with the right approach can be acquired by anyone.

Calculus classes are more passable than programming classes, and the claim you're making is just pulled out of your hat, made in such a way that you could never be found wrong, because no matter what you could just say you're not taking the right approach.

Reply on news.ycombinator.com