Thanks for the feedback. Some interesting responses. Firstly, the title of the forum post was slightly provocative on purpose to promote some discussion, and secondly, this forum was likely to produce the harshest critics!
What I find interesting about the responses I've read above is that lots of people say "you have consolidated/cut down the list too much" or "it doesn't feel right to have lost that many cars", I can't see that anybody has pointed out any specific examples of over-consolidation. Something like "your entire article is of questionable accuracy" only becomes useful when a "because" is added!
So I justified this with "434 distinct individual cars that are not just re-skins of the same car, or the same car with a slight difference (1.5-litre instead of 1.3-litre)." I was careful not to say duplicates, so it's interesting people have interpreted it that way.
I like Bigbazz's point about having such a huge "library" of cars so that you're more likely to find the specific car your dad might have owned while growing up. Unfortunately, because there's a 2-in-3 chance it'll be low-poly, low-quality in GT6, I think it might ruin my memories, rather than make them fonder! (Although this might not be the same for everybody).
Also Bigbazz touched upon rFactor 2 actually looking better than GT5 these days. I suppose I didn't think I could justify the space in the article, but with rFactor 2, it might well be that the "official" cars and tracks look pretty good, but the user-made ones are generally pretty poor in my experience, so the "overall quality" in my mind is a conglomeration of the official and fan-made, which might not be objective enough.
One final point I left out of the article because I didn't have the data to justify it: how many of the 1197 vehicles have close to realistic physics? It seems hard to imagine that it's possible to accurately model the differences in handling and response between a 2005 Suzuki Swift and a 2007 model. Was each model dismantled in order to determine things like chassis rigidity, or driven hard to measure things like torque steer? Genuine questions, as I don't know!
Do those two cars just handle in the same way, and look the same? If that's the case, then are they really two cars, or would it be better, like Wardez suggests, to just have a "Suzuki Swift" and then you can pick the year. But then if they are exactly the same, why not put in Suzuki Swifts from 2004, 2006, 2008, etc since these are all the same platform with minimal differences?
Also would anyone seriously consider buying a game that contained only 81 Skylines, if there was an equivalent game right next to it on the shelf that had 81 cars, all from various manufacturers?