While I was looking for a job, one of the places I applied to asked me to do a selection test at testdome.com .

First of all, the job was asking for a C# developer and 50% of the questions were pure JavaScript and 50% pure SQL.

I found this ridiculous. The nature of the JS questions were silly. They have you quickly hack up functions and there are corner cases and you either get those corner cases or you don't... Maybe it's just me but, who doesn't use JQuery? Who on earth resorts to pure JS functions to get work done? Also, even if you did not get all the silly corner cases that would never happen with JQuery, the quality of your code is not taken into account.

As for the SQL, I felt the questions were intentionally tricky. But once again, what sane developer DOES NOT use an ORM? LINQ, Rails, anyone?

One of the questions gave me 20 minutes to turn a traditional table into one made from divs and classes. My solution gave the exact output they were looking for, but the system did not see the solution as valid at all.

I wound up failing the selection test and my interview was canceled with them.

The job I wound up getting, which I'm very happy at, gave me 30 minutes to answer theory questions on paper such as:

What is the factory design pattern? Explain
What is MVC?
What is refactoring?
Write the class methods for a binary search tree (no implementation)
Explain inheritance vs composition; when should one be used over the other.

A bunch of questions like that. I aced it and got the job. At this job, I'm working with high level frameworks and ORM's like any sane developer.

All this to say, what are your thoughts on these kinds of 'selection tests' (like the one from testdome)? Do you think it really gets the best programmers?

For me personally, I may not know all the corner cases for working with pure JS, but my code is really clean, clear, well tested, well documented, my architecture designs are very modular, abstract, generic, and I'm really proficient with frameworks like Rails/ LINQ, with Backbone.js, etc... the tools to actually build software. The fact that I failed the selection test I feel is an indicator that the selection tests are bad.

Then their company has a systemic problem of people relying more on "standards" than "knowledge" and you're likely dodging a bullet!

Plenty, actually.

In my case, we started out using Hibernate, but switched to a purely hand-written SQL framework because it forces us to actually think about what we're doing to the DB. You can create some pretty sloppy hierarchies if you're not careful.

And now an anecdote. Using an ORM, a client I was once working with ended up with a query that used (I believe) 63 joins, which exceeded the maximum that MySQL was capable of. I don't recall the solution, but I imagine it wasn't a fun refactor.

Edit: When working under the pressure of an interview, it's hard to sometimes catch all the corner cases. But not being able to identify them is a red flag. Software is always filled with unforeseen corner cases: we call them "bugs." While I didn't personally review the tests you took, I could envision a technical interview including questions with obvious corner cases, to see if the applicant accounts for them.

If I ran a company (and I intend to), I would focus much more resources on hiring the right people first, instead of allowing crappy workers to draw down the morale of my key people until a laggy, slow, easily gamed HR system eventually kicks them out.