Comment by fredkbloggs | Hacker News Reader

Comment by fredkbloggs | Hacker News Reader

The author was doing great right up until the last paragraph. His proposed solution is silly because it makes no distinction between jobs for which a university education is essential and those for which it is irrelevant. Then there is also the large grey area in between, in which a university education in certain fields might be useful but not essential. What he would do is abolish the right of those professions in which such an education is essential from applying the same standard he lauds earlier: that physicians must attend medical school. For engineers, a university education with a specialty in engineering is exactly what medical school is to physicians.

Many people here believe that a university education is not required to be a computer programmer, and I agree. But programmer is a vocation, like plumber or firefighter; it requires specialized skills and experience to do well but does not require a university education. If you spent 3 months learning one language and one or two common frameworks for making e-commerce sites, you are reasonably qualified to work on e-commerce sites for companies that use that language and framework. However, being a programmer and being an engineer are not the same; an engineer has been educated both more broadly and more deeply, and should be capable of solving larger and more difficult problems, designing whole systems, and in general should have the education and experience to understand new problems more thoroughly. It's the difference between the architect who designs a building (or one of its systems) and the construction crew who erects it. Both are essential, but they have (properly) different qualifications.

There are many other fields where some sort of education is required. I agree that many of those fields conveniently piggyback on university education (usually with a particular field of study requirement) when other similar mechanisms might do as well. But the notion that such an education is universally irrelevant is absurd.

If the author wants people without university education to be paid as much as people with it, the solution is not legislation. Instead, he should hire them.

His proposed solution is silly because it makes no distinction between jobs for which a university education is essential and those for which it is irrelevant.

His solution is to evaluate people as people, not by external indicators. If it so difficult to tell if someone has a university education or not without having to explicitly ask, then perhaps it isn't as essential as you seem to think?

> If it so difficult to tell if someone has a university education or not without having to explicitly ask, then perhaps it isn't as essential as you seem to think?

It is trivially easy to think up questions that any Computer Science student from a non-crappy university will be able to answer, but someone without a degree probably will likely not be able to answer. Just pick a particularly challenging problem off of an Algorithms or Linear Algebra final and hand them a textbook and an hour.

The question is how to ask questions that matter for a particular job. If the job is hacking out an e-commerce site, it's really easy. If the job is designing a distributed system for real-time processing of data at 100s of GB/s of data that nobody has tried processing at that speed before, the single set of questions that tell you whether this person is competent is far less obvious. What you really need is to engage with the person in weeks or months worth of examinations on various subjects. Fortunately, university courses have midterms and finals that do exactly this.

"If the job is designing a distributed system for real-time processing of data at 100s of GB/s of data that nobody has tried processing at that speed before, the single set of questions that tell you whether this person is competent is far less obvious. ... Fortunately, university courses have midterms and finals that do exactly this."

But solving problems that nobody has solved before is the kind of job that's usually filled by senior developers, not by recent graduates. By the time you have 5-10 years of work experience, the things you learned during your working career probably play a much more important role than the stuff you learned in college. If I've spent the last ten years actually designing distributed systems, why does it matter whether I've gone to college or not?

> But solving problems that nobody has solved before is the kind of job that's usually filled by senior developers

> By the time you have 5-10 years of work experience, the things you learned during your working career probably play a much more important role than the stuff you learned in college

I submit these two quotes are not contradictions, but are at least in tension.

In general, 5 years hacking out code doesn't prepare you to write code that's mathematics heavy or design systems that require careful mathematical analysis.

> If I've spent the last ten years actually designing distributed systems, why does it matter whether I've gone to college or not?

It doesn't. But I bet I can pay a fresh CMU/MIT grad less, and that they'll be able to figure things out.

"But I bet I can pay a fresh CMU/MIT grad less, and that they'll be able to figure things out."

Sure, but it could take them much longer to figure it out. College work doesn't really prepare people to work on cutting-edge problems that nobody has solved before. Problems assigned to undergrad CS majors are generally straightforward applications of what was taught in the course, and the professors already know the solutions and how much work they involve (e.g., an assignment due in two weeks is known to be possible to complete in two weeks of effort).

"In general, 5 years hacking out code doesn't prepare you to write code that's mathematics heavy or design systems that require careful mathematical analysis."

How many CS grads have had experience with mathematically oriented code? The only math I remember doing in my CS classes was in theoretical courses like Algorithms, and that mostly involved proving things rather than writing code.

If you are able to trivially weed out those who do not have a CS degree, and having a CS degree carries with it the necessary skills to do an previously unattempted processing task, then doesn't the former naturally solve the latter?

The author does not seem to have a problem with testing for the presence of skills, like your algorithm question, but making an assumption that someone does not have skills based on their race, gender, sexual preference, or credentials is where he takes issue.

> but making an assumption that someone does not have skills based on their race, gender, sexual preference, or credentials is where he takes issue.

This is answered entirely in my original post. Adequately testing for the presence of skills might take weeks or months of testing, which is a prohibitive cost. Fortunately, a huge portion of the potential employees I have to interview already took those tests throughout their coursework, and I can pretty well trust the quality of the tests they took based on the institution they went to (unlike SAT/credential/GRE-style tests, which are too easy to game). Why should I blow thousands of dollars in my time testing the others? How is that fair to me? And why would I assume someone who never took a math course after highschool is going to be able to work on my linear-algebra-heavy software library?

Also, lopping university degrees and other credentials into a list of protected classes is insulting. A degree is something that is earned and requires time and effort on the part of the student. It's expensive sometimes, but it's also a lot of work (assuming it's a good university)

> Adequately testing for the presence of skills might take weeks or months of testing

So it is not trivial to test for the presence of a CS degree? I'm left confused as this contradicts your earlier post.

> A degree is something that is earned and requires time and effort on the part of the student.

On the flip side, is it not insulting to someone who put in that time and effort via non-traditional means to be passed over because you metaphorically "didn't like the color of their skin"?

> So it is not trivial to test for the presence of a CS degree?

It is trivial to test for the presence of a CS degree. The bulk of a CS degree can be a necessary but not sufficient qualification for a job. Testing for just the extra stuff is far easier than testing for everything.

This isn't rocket science.

(And no, discriminating based on credentials is not tantamount to racism. I would have hopes the problems with this analogy were fairly obvious...)

I'm sure I'm just misunderstanding you, but the logic doesn't seem to follow. If you can trivially test for a CS degree then you already have all the other testing that comes with a CS degree by association. Why would you take weeks or months to test everything in a CS degree when you already know a CS degree is present via your trivial test?

> And no, discriminating based on credentials is not tantamount to racism.

You are, however, judging someone's abilities on their outward appearance even though in our hypothetical situation the person is equally skilled as anyone else. It may not be racism exactly, but still morally wrong for many reasons. That should be fairly obvious.

@randomdata -- "trivial" isn't "free". The test I proposed adds an extra hour of in-person interview time per candidate, and also substantially increases the number of viable candidates.

> It may not be racism exactly, but still morally wrong for many reasons. That should be fairly obvious.

Comparing discriminating based upon whether someone did four years of audited work with racism is... a somewhat extreme position, to say the least.

You may not think that college is "real work". You're welcome to your opinion. But I assure you that a university degree at many universities does require a lot of work, and that this work is difficult enough and audited well enough that it warrants consideration in hiring decisions.

A degree is NOT something anyone is born with. It is something many people work very hard to earn.

To put a finer point on it: is it OK to discriminate based upon prior industry experience? Because I consider a degree from a top uni better preparation "work" preparation for many jobs than X years writing web apps.

Is it OK to discriminate against doctors who don't have MDs?

Or am I supposed to give everyone who submits a job application the benefit of an interview, and sink my business in the process?

> A degree is NOT something anyone is born with. It is something many people work very hard to earn.

But, again, you are automatically discounting everyone who worked equally hard to acquire their skills just because they don't have the right piece of paper. That is exactly the same as rejecting a female candidate on the assumption that because she is female, she probably doesn't have the necessary skills – and statistically speaking, in this case you'd be right – the number of female CS graduates is significantly lower than men.

So, if you're just playing the numbers game like you suggested at the beginning of your post, then why not reject all women?

> then why not reject all women?

Because there is often a CAUSAL link between the degree and the set of qualifications relevant and related to the job at hand. The link between skill and gender is not CAUSAL.

You're pretending like there's no causal link between education and capability. You are wrong. Period. Hiring someone based upon degree is absolutely no different from hiring someone based upon whether they have prior work experience.

Need someone to work on a team? Deliver some production-quality CRUD? You could test these directly for every single candidate in an interview. Or you could hire someone with prior web dev experience and a letter from the previous employer or a portfolio that directly and immediately witnesses the requisite knowledge and skills. Everyone does the latter (or uses some other signal to weed out who they interview) because the former is a stupid waste of resources.

Need someone who can model a system using some ODEs (this afternoon, not in 6 weeks after wasting you money taking a MOOC on company time)? Making tweaks to a library for processing large matrices? You could give every single candidate a pop quiz on ODEs or Linear Algebra. Or you could hire someone with a degree and high grades in those courses that directly and immediately witnesses for requisite knowledge.

Another way of saying this: Degrees are direct witnesses of relevant skills, not merely signals or correlated values.

Of course there are other direct witnesses. But there is never anything remotely morally questionable about taking into account the presence or absence of any direct witness of capability during hiring. Including degrees.

> The link between skill and gender is not CAUSAL.

I'm not sure it is relevant to the discussion. The question was asked in the context of your concern about the overhead of interviewing people. You are using a degree as a simple filter, and gender is equally applicable in that regard. Statistically, for our aforementioned example, a suitable candidate will be found in the male population of applicants due to the reasons we already discussed.

So, why bother wasting your time and associated costs also interviewing women when you can be reasonably confident that the male candidates will provide someone suitable? It is very possible that a great employee lies in that group of women, and it is very possible that a great employee lies in those without degrees, but you have made it clear that you don't have time to evaluate them all.

> You're pretending like there's no causal link between education and capability. You are wrong. Period.

I'm getting a distinct impression that you haven't fully understood my comments. That may very well be my failing, but this becomes a completely pointless discussion if that persists.

> Hiring someone based upon degree is absolutely no different from hiring someone based upon whether they have prior work experience.

Absolutely not true. What you are essentially saying is that even among those who have prior experience, you will only hire those whose experience came from Google or Microsoft. There are a number of practical reasons why a perfectly suitable candidate is unable to work at either of those companies before working for you, even though they do come with relevant experience from somewhere else.

I understand from your point of view why hiring from only past Google and Microsoft employees is desirable, but what about the people who were unable to work at those places? You really don't see the problem with that?

> You really don't see the problem with that?

I don't see a moral or ethical problem with it, no.

And the democratic consensus in most western countries agrees.

Choosing one applicant over another because of prior work history or education is just making informed decisions. Choosing on applicant over another because of their gender is illegal discrimination.

I understand you feel like people who don't have degrees deserve equal footing in the job market. But they don't, and business don't have any moral, ethical, or legal obligation to forego their own best interest and take bets on higher risk hires.

Basically your entire argument boils down to: "It is the status quo", which is not a very good reason for why we should avoid change.

There was a time when it was considered moral, ethical, and legal to hire based on gender and race. Can we assume you have no qualms going back to that so long as the democratic consensus agrees?

No, that is not my entire argument. (However, ASIDE: when every advanced nation disagrees with your view on morality or pragmatism, it does not mean you a wrong, but it does mean you should closely examine your line of reasoning for both pragmatism and consistency with reasonable and well-grounded moral standards. Of course the democratic consensus of all of these countries can be wrong, but it's worth taking pause. That's all. END ASIDE.)

Here is the argument that I have been expressing in various forms and you don't seem to understand or provide cogent counter-points to:

PRAGMATIC CLAIM: Employers are allowed to "discriminate"/select based upon non-intrinsic signals of characteristics relevant to the job at hand.

WARRANT: This is necessary for society/businesses to function because otherwise, it would be prohibitively expensive to hire people using anything other than coin-flips.

MORAL STANDARD I: In order to use a signal as the basis for a hiring decision it should be both a) non-intrinsic, and b) signal something relevant to the task at hand.

EDUCATION MEETS THIS MORAL STANDARD:

a) CLAIM: Education is non-intrinsic

WARRANT: It is something you seek out and work to earn, not something you are born with. (The issue of ensuring universal access is entirely separable (and something I support, btw).)

b) CLAIM: Education signals something relevant to the task at hand

WARRANT 1: Technical background is assured through testing in relevant coursework -- testing that you would have to (at great and impractical expense to the business) reproduce in order to make informed hiring decisions.

WARRANT 2: Education can also serve as a signal of necessary problem solving ability, communication skills, etc.

MORAL STANDARD II: Any moral constraints placed on the set of available signals should not make it prohibitively expensive for business to function.

EDUCATION MEETS MORAL STANDARD II: Because it's a cheap signal to use as an initial filter.

BANNING THE USE OF EDUCATION WOULD FAIL TO MEET MORAL STANDARD II: It would require companies to essentially require a comprehensive examination that is well-proctored etc, which is a pretty huge burden for the company and -- besides -- basically would amount to credentialing anyways because any reasonable firm would outsource this.

Other relevant claims:

CONSISTENCY CLAIM: There exist signals that meet both standards.

WARRANT: Education can be used to select a small number of applicants, and the cost for choosing among those should be marginal wrt the value-added by the employee.

CLAIM: These moral standards preclude the use of race/sex/age discrimination.

WARRANT: these are all intrinsic features that are explicitly forbidden

I also think that there is no practical way his solution could work, at least not with the laws the way they are. How would you ask people to explain gaps in their work history? Are you no longer allowed to consider why someone was unemployed for 4 or 5 consecutive years? Additionally, I don't think you can easily do general intelligence testing for a job. You have to be able to show that every question relates to job performance and that it does not cause a disproportionate impact on the hiring of people from protected classes.

Reply on news.ycombinator.com