Pretty sure that's exactly what Tennenbaum was doing. Rather than admit, "the test didn't produce a result," he'd just make it up. Asking for information about period cycles and intercourse sounds like he was using that information to try to guess, with no more accuracy than the mother could have done by herself. They weren't even collecting enough blood for an accurate test, if he even really ran them. I bet it comes down to a combination of greed and hubris.

Or regulate at all...

Associate Prof. Ma’n Zawati, research director for McGill University’s Centre of Genomics and Policy in Montreal, says private commercial DNA laboratories don’t need licences to operate and sell services.

😲 That's some pretty serious caveat emptor there! I'm not giving my dna so any random weasel in a suit.

Generally, how DNA tests work is by selecting points in the DNA and comparing the two samples. The more points you compare, the more likely you have a true match. If you don't compare enough points, or pick places in the DNA that are unlikely to have much variability across the population, you'll get all matches on those points and say it's a match. For paternity testing, you're looking for ~50% matches.

Though, in this case, it does look like they were just making stuff up:

Richot said she was coached to ask women seeking prenatal paternity test kits about times in their menstrual cycles and the dates they had intercourse with different men — information that is useless for a DNA test.

Staff put the dates into an online ovulation calendar to narrow down the possible biological father, she said. Richot then entered the information into a form that went to Tenenbaum for signoff.

"[Tenenbaum] would always make a comment like: 'It's definitely this one [the biological father]. It's this one, it's got to be this one,'" said Richot.

since people tend to assume paternity tests are infallible

This specific instance is particularly egregious, but...

They shouldn't be doing this regardless. Most medical tests have a false positive rate, and I suspect nearly all paternity tests do as well. When interpreting medical results having a decent understanding of Bayesian statistics is very useful. If you don't, you should be asking your doctor to help you interpret results.