But it’s like this: I felt that this particular argument hasn’t received much notice, and I thought people would find it interesting to think about.
It actually puts a legal requirement to do the opposite - to hire the best candidate regardless of their race, national origin or their interest in yiffing at FurrCon.
At least in the US, (and I strongly suspect in Canada as well) it's up to a jury to decide if the given rationale is genuine or an excuse to cover bigotry. Anyone who knows anything about how to do the job is likely to get blocked at jury-selection. Imagine explaining to 12 literature majors why cubic-time lookup is bad, while opposing council wants to know how often you use key-value stores in your day-to-day work.
Not always. For example if you somehow learn that a candidate plans to take several months parental leave in the near future, they can't possibly be the best candidate, but you're forbidden from taking that into account.
If you believe for instance that
1) it takes 3 months of ramp-up time to do an effective job 2) after 3 months off the job, you need to ramp up again 3) the employee will leave in two months 4) the employee will be gone for twelve months
You could conclude that the next 17 months of this employee's tenure at the company will be ineffective.
You could contrast this with an employee that works straight through and will give you 14 effective months.
You might get an exceptional candidate taking the leave, but they have to make a huge impact in a short time to be better for the company on balance than an extra year of labor from a roughly equivalent candidate.
Further, if it's not plainly obvious that the practice is wrong to someone, it seems like stating the fact probably won't do much to convince them, so maybe alternative arguments might be more productive.
Or to take the approach advocated by Milton Friedman and pass a law to constrain their behavior. The infinite loop is introduced when corporations can influence which laws are passed...
In an industry (and this includes ours too, techies, don't be deluded into thinking otherwise) that's driven by the bottom line, practically without morality, perhaps the best argument you can make is something like this.
The point he makes is - "this policy stops us writing quality software ". Now assuming writing quality software is a benefit to the company and it's shareholders this point has a lot more weight than his persona moral scruples.