Hacker News Reader

Comment by Moshe_Silnorin | Hacker News Reader

>in 2015, then watched as more than 900 applications rolled in. From this pool, they chose 11 former miners who scored highest on a coding aptitude test. Two years later, in an old Coca-Cola factory by the Big Sandy river, nine men and one woman remain

They gave people an IQ test and took the top percentile of a group that, due to their interest, was already slightly selected for intellectual ability. This is not proof that anyone can program.

Did you even read the article? It is in accordance with the unfortunate fact that programming is a high IQ job that the majority of people are not capable of becoming proficient in. Though for many here this is not unfortunate, as it is the reason wages remain high despite a lack of licensure.

Humans differ in cognitive ability and most of this difference is genetic. Modern psychometrics and common sence agree with this. We cannot build a better word on false but pleasing ideological assumptions.

View on HN

Thanks for saying this. To me, in a strange way, it's a little arrogant to think I am being generous by saying that anyone can code. It feels like I'm engaging in false humility and I am putting myself above them in the first place to confer this "honor" upon them. I'm also subtly devaluing the many other things they probably can do.

But, maybe I'm over-thinking that part. Either way, I have witnessed first hand that not everyone can code, and we have probably all seen this. I mean, it's a vague statement to begin with. Sure, you can probably show most people how to write a for-loop, but that's a far cry from them being a proficient, professional coder, which is really what's implied by the phrase "learning to code" if used for any practical purpose.

And, you can't just chalk it up to bad teaching. I've seen many people, in school and at work, who just didn't have the thinking style. In school, so many people couldn't turn the corner, then switched majors and did perfectly well. They weren't unintelligent. Coding just never clicked. And, as a tutor, I really witnessed it up-close. For some, I was able to guide them to finding the solution, giving them just enough to turn on their inner coder and help introduce them to that part of their brain.

But, for others, no matter what I did, it just didn't click. I could show them pieces at best, but they could never seem to put it all together. Felt like it was just something about the way their minds worked.

And, there are some who somehow made it through school, but struggle professionally and it shows in the quality of their code. The idea that we can make anyone "good enough" is unrealistic.

Bad code costs money and it's economically no different from any other skilled field, wherein firms that employ highly skilled resources are at an advantage over those who employ less proficient resources. And, this is not just true for Google.

Agreed 100%.

My experience is that being a good programmer is more than a matter of intelligence. It's a certain thinking style that lends itself well to organizing data flows and presentation. It's a certain ability to focus on abstract structures and keep chains of relationships in the mind while envisioning how to manipulate them. And a lots of otherwise smart people just don't think like this natively and "teaching" them is both painful for everyone and has limited effects.

I bet most of us have met perfectly smart people who just didn't click with coding. Just not a good fit for their thinking style. Even if they nominally want to be programmers for the money or cool factor or whatever. Even if they take a bunch of classes. They just never quite get there.

And square pegs beating themselves into round holes usually does indeed cost more than it produces. People should play to their strengths instead of chasing external expectations. It's better for them and for society in my estimation. But certainly if people want to give programming a try they should. And at least "some" experience in programming is better than none. But most people can't/won't become professionals.

Exactly. Really reminds me of proofs in geometry. You amass this knowledge or ruleset, then apply that knowledge in successive steps to solve the proof.

But, the bigger part is being able to step back and "see" the path; to pull from what you know to approach the problem and formulate a solution.

Seems like there were three groups: those who just couldn't get it; those who wrestled and muddled their way through; and those who just kind of got it, almost intuitively. The latter is a minority, and I would bet that an outsized percentage of those people really can learn to code.

To me, the process is more suggestive of the idea that the resources likely to produce success when training people to code are scarce and access to those resources is therefore limited using highly selective processes to guard against allocating resources to false positives.

I suspect that the process selected 11 people because there were funds to educate ~11 people and that ~11 would have been enrolled in the program if the applicant population was significantly different regardless of that difference being a left or right skew.

The resources are not scarce; they are free online.

There are resources for many endeavors that greatly increase the likelihood of success. It might be a Harvard education. It might be YC's startup school. It might be a retraining program in Kentucky. All are scarce and therefore selective.

One of the things each offers is sustained direct access to other people's time and in particular sustained direct access to expert's time. That's the difference between a personally tailored answer and closed as duplicate on StackOverflow. It's the difference between here's the answer and ten other things I can tell you about and here's someone else's question and several divergent answers to your duplicate question which might not have been the same.

I run an online code bootcamp, and I think most people assume that learning resources being available is all that's needed. That's true for some super dedicated people, but for most it's not enough.

For a while I thought the biggest value we provided as a bootcamp was direction and on-demand expert help. But now I don't think that's true. I think the biggest value we provide is that we make everyone show up at the same time every day and work until some other time, and we help direct what they should be working on during that time.

In short, it's social pressure, discipline, and scheduling. Those are everything.

Look at Thinkful's "part-time self directed" bootcamp - all the material is there, there's a mentor to help you once/week for a few mins, you pay $8,500 to do it - people are obviously dedicated to learning when they start. And the end graduation rate? ~31%.

I know another bootcamp that decided to just record their bootcamp and put it online, and charge you $2,000 for access to the videos. 28 people signed up and paid the $2,000, and one person watched past the first lesson.

I would guess that the percentage of people who can discipline themselves enough to learn on their own while balancing other responsibilities and not paying $2,000-8,000 is much, much smaller.

Think back to when you were in college - say it's a Calculus class. What do you think the graduation rate of that class would be if you gave everyone a Calculus book and said, "Fit it in when you can!" I'd guess it would approach zero except for a few superstars.

The same thing seems to be happening in programming. In 2017 the thing holding you back from learning something isn't resources. It's direction and discipline.

I missed the IQ test part. All I read was coding aptitude test, whatever the hell that is.

Sure we're all different and maybe 95% is high but its at least 50%. One out of every two people I meet I am confident they could become proficient at programming given a solid learning environment.

A programming aptitude test is just an IQ test with another name.

>One out of every two people I meet I am confident they could become proficient at programming given a solid learning environment.

That's quite possible, even if what I say is true. People tend to mix with people on their own intellectual level. The people you talk to enough to get a good estimate of their intelligence are unlikely to be randomly selected from the population.

If you believe average people can become good programmers, this implies you could make billions converting 100 IQ people into programmers and then undercutting everyone else. I predict you will fail, unless you figure out a way to vastly deskill the art of programming.

An alternative way to prevent deskilling programming is to make the tools required to program such that learning to use them becomes mostly a matter of oral tradition rather than accessible. This might (or might not) be happening.

For example, the evolution of Javascript from a low status programming language to a high status one has paralleled a shift from a stable core language and a well documented API to the DOM to a combinatorial explosion of changing frameworks and complex development toolchains. Learning some combination gulp and babel and redux and angular(2) and webgl and ten other things is more a matter of access to people who will patiently answer questions than something a person can learn without working on a team.

I think your learning style is leaking through here. While not a JavaScript expert (who designs a language with whitespace/formatting not being a first class concern?) I had no trouble picking it up from the docs and random blog posts.

I'm a learn-by-doing type though, YMMV.

I agree people's mileage varies. My gut is that programming languages have generally evolved in ways that widen that variation and that the explosion in the number of people who are coding due to increased access masks much of that change.

I once worked with a reasonably productive developer who said his IQ was in the mid 80s. I think whatever test he took misjudged him, but it does suggest that, while programming skill might be rare, and might correlate with IQ, it's not a universal law that only high IQ people can become programmers. Also important to remember that some programming is about as complex as simple electrical work, while some is significantly more challenging. It would be unjust to deny people the opportunity to do the former because they can't handle the latter.

I studied math & computer science at university (15 years ago) that was quite selective in CS admissions (probably filtering out ~80% of high-school graduates).

At least half of these already self-selected and 80% filtered students were not really able to program (and eventually flunked or transferred to other departments - although some people who could program flunked and some who couldn't graduated, I think the two groups were quite well correlated).

I'm sure you can do better at teaching (and filtering) than the CS program at my university, but if such a focused group had a 50% failure rate I'm sure the general population is at the very least not better (remember that it's not just about 'raw' ability but also about interest and motivation).

It's not a proof that anyone can program, but neither is it a proof of the opposite.

You don't need to be the top 1 percentile to make CRUD apps, they chose the best because they had the first pick, not necessarily because they needed to.

I think the idea is that many people will never be able to make a stable crud app in a reasonable timeframe, no matter how hard you try to retrain them. This isn't the 1%, it's just that the majority of programmers are more proficient than the general public in some way. Sure that's hundreds of millions of people, but it's not everyone.

It's quite possible that you need to be the top 10 percentile to sling together CRUD apps. I think trying to minimize the difficulty of what we do is a little insulting. Even the dumbest software developers I've met are sharper than the average bear.

I mean, get out in the world and see what most people are like.

Seems like a better filter for attitude than aptitude.

Reply on news.ycombinator.com