Slashdot Mirror

The report says that it "is based on a sample of more than 36,800 students from 500+ colleges across India". It doesn't say what degree they're in, how much experience they have, or a variety of other factors to conclude the 36% of Indian engineers can't code. This is just a racist hit job piece.

I looked into it and based on another study they did yeah, they tested all engineers. It looks like they wanted scary numbers so employers would be alarmed into buying licenses to administer the test, and universities would be alarmed into buying licenses and materials to teach the test. Because this looks like a seriously fudged research-design.

Different but similar study by the same company it looks like they are testing engineers from all fields just so they can publish an alarmist report conveniently failing to mention that it's just SHOCKING that a mechanical engineer wouldn't know Object Oriented programming! /s These guys are selling the very test they studied. This is spam that got picked up because it looks legit and regular news (and slashdot editors) don't bother to investigate when given a chance to print outrage-pr0n. It's scaring students, universities, and employers into licensing the test or test-training resources.

...OK, I found a link to a different study by the same company with similar themes. It looks like parent and my suspicions were warranted. So in the case I found, Aspiring Minds did a study where they gave their own test, the computer programming portion of the AMCAT (A product that they sell), to engineers of _all_ fields. And they drew similar alarmist conclusions to what we're seeing in TFA. I'd love to have more information on the actual study, but, for some reason, I can't seem to even find so much as a citation, let alone an abstract of this study anywhere...so, sounds like an private press-release. Meanwhile, The link I did find has no authors attributed to it, it is not published in a proper journal, and there's no evidence of peer review.

TFA is about giving a somewhat different coding assessment they call "Automata". Whereas their "Computer Programming" test is language agnostic, this one is language specific. Supposedly the test takers do indeed get access to a compiler and a "simulated integrated environment" (and who's to say if that thing is any good?), but at least it sounds like you're permitted to compile and execute test-cases before submitting your answer to a problem....Sounds to me like a test where unfamiliarity with the test is seriously gonna hose you. And if the study isn't gonna affect your grades, then what motivation do you have to seriously try your best on it??

This is trash. They aren't gonna be able to filter out biases in test design, test execution, or in making conclusions. Why on earth would you bother to test non-computer engineers on programming skills? What knowledge do you seriously gain from giving a bunch of students a test on topics that their field of study usually doesn't care about and acting surprised when they fail it? From what I can see, this is a company that is attempting to sell its services to companies doing interviews. So this is a FUD paper: "You'd better purchase the license to give our test to your prospective hires or who knows what horrible employee you'll get! And hey, universities, maybe if you worked with us more, these scary numbers wouldn't happen quite so much!"

This story is only spreading because the results sound scary and because people just love the narrative that offshore software development is a bad idea. And IMHO, if you aren't good at building requirements and hashing them out remotely, then I think it is indeed either a bad idea or a pretty painful ordeal. But it's not because India is a nation of incompetent software engineers! It's because it's really tricky to define what you want in a fire-and-forget manner.

I was somewhat skeptical of the numbers as well, and found a previous year's version of the same survey: 2016 Report (PDF) It seems that part of the problem is that it looks at any type of engineer, whether computer, electrical, software, or mechanical. It also measures employability in fields such as civil engineering, chemical engineering, and other fields that have nothing to do with software development. The numbers for some of those fields are higher than the number quoted in the summary, which leads me to believe that a reasonably chunk of the engineers surveyed have no desire to program at all or pursue a career in it.

There are some other interesting figures in the report, but it's quite large. Seems like this is another case of a reporter not understanding a study and making a bad headline.