My claim is we could mostly rule that out our batch of universes contains life-poor universes if we could randomly sample a universe and see if it has life.
That matches up to my position though, as I also think that if you find life in a universe, chances are that it isn't a life-only-once-poor universe, which is a tiny subset of life-poor universes.
I don't see the basis on how you can make that claim. We don't have a general handle on what kinds of universes are out there with what frequency. You can make guesses but that's assumption laden and shouldn't be taken as more than a guess, you'll always be able to make counter assumptions to skew it in a different direction and I see no firm method to arbitrate the claims.
Like you said yourself, chances are we are not in a life-poor universe, but instead probably one of the many other kinds. By implication a subset of life-poor universes (which life-only-once is) is going to be even less probable.
Let me make an example where this type of reasoning falls apart: Suppose you have two bags of marbles with 10 marbles each. One of them has 10 white marbles, the other one only 1. Now you chose one bag at random, take one marble out and it is white. The probability that the bag you chose was the one with 10 white marbles is 10/11 = 91%. But now consider the situation when there is one bag with 10 white marbles and 1000 bags with 1 white marble. Again you chose one bag at random and take out a white marble. In this case, because there are so many of the 1-white-marble-bags, the probability that you chose the one with 10 white marbles is only 10/1010 = 1%.
The issue I have with this is that you are setting up the universes ahead of time. You have to assume that you just don't know what sort of distribution it will be, so you can't just pretend you have 9 bags with only 1 white marble in it. That is just one of the many possible combinations of bags possible, you have to consider all the other scenarios as possibilities.
The way the bags are constructed, each time you decide whether you put a black or white marble in a bag, you roll a die. If a certain number appears, you put a white marble in, otherwise you put a black marble in the bag.
A situation in which 9 bags will only have 1 marble and all black marbles, and 1 bag containing all white marbles, is not very probable. Statistically speaking when constructing your bags like this, even if you don't know the probabilities, you will not expect to end up with such a situation after all the dice have been rolled. It's of course possible but if you repeat this exercise 100 times, you will not expect 1 of those situations to result in such a distribution of marbles.
The essence of my argument is that statistically speaking ending up with 1 white marble and 5.6 x 10^20 black marbles is not as likely as all the other possible scenarios, of which there are many. It's simply because "just 1 white marble" is one tiny data point, while all the other possible scenarios overwhelm it to a rather large degree. If you were a betting man, you would not bet on a bag containing only 1 white marble, you would assume that the weight of all the other scenarios is a far more likely situation