Boys, could you ever fancy a plain/unattractive girl?

Attractiveness is not just reference to looks, it is also in reference to personality. I mean girls go crazy for Benedict Cumberbatch as Sherlock Holmes and Matt Smith as Doctor Who and neither of them are conventionally good looking. So if girls can fancy not-very-good-looking guys then it could work vice versa too.

No if I was to date someone i would want a girl with a good personality whom i also find attractive, otherwise we would just be good friends.

If you had ten people and graded them in attractiveness from one to ten, and the person who was a seven improved themselves physically up to a nine, then the people who were eight and nine would shift down. To suggest otherwise is nonsense. Even in a room of supermodels you would find one more attractive than another and grade them from one to ten (i.e. normative rather than criterion referenced assessment).

The fact you see lots of good looking women shows you have a limited sample size (i.e. young girls are statistically the best looking of the whole population).

[…] To use another example, if I had 10 girls who spanned absolutely hideous (1) to utter perfection (10) in roughly equal "steps", then replaced all but 4 and 10 with more 1s, I wouldn't suddenly find 4 any more attractive (even though on your scale she'd now be a 9). I may be forced to lower my standards to improve my chances of getting any girl (if the 10 was simply out my reach), but the 4 would look no better than she had originally. […]

I'm going to be straight with you.

Hell. No.

Tbh I'd go with a girl who's less attractive. There's more certainty shall I say? Less likely to be deceived.

You have missed the point. The values of the eight new girls apply only to their previous scales (i.e. they were all considered the ugliest). These values are arbitrary so you saying one is absolutely hideous and ten utter perfection is actually irrelevant.

For example, your first scale could be the ten hottest girls at your school. All the girls are likely to be physically attractive. But ultimately you will be able to grade them from one to ten in attractiveness through statistical sampling. Now if you take the 'hottest' girl out (i.e. ten) and replace her with a number one from another scale, and I think we agree here, that does not mean number nine is now the hottest girl and the new number one continues to be the ugliest. It is a completely new scale which needs to regraded. It might just so happen that the new girl is Adriana Lima and she was considered the ugliest from on a scale of the hottest supermodels in the world, and would likely be the number ten in this new scale. The presence of Lima would make the other girls less attractive; even more so than their scale itself suggests because it is not criterion referenced. In this case, and I have used this example deliberately, the number four – despite having exactly the same rating as before – would actually be uglier now than she had been before. In fact, assuming there was a comparable scale of attractive men who had been there for both scales and would have been paired off equally (i.e. woman ten with man ten, woman nine with man nine), the number four girl would probably lower her standards now because of Lima's presence (i.e. men one to four). In contrast, male number four would probably raise his standards in response because Lima has increased the unseen mean in this new scale and it is very likely that pairings which not match up as well as the first time, especially if you made it a pressure situation (i.e. limited number of girls to ask to go to a dance).