So does this then boil down to either all similarities being a matter of accident vs all differences being a matter of accident?

If so, less Metaphysical violence would be done by extreme realism as there doesn't seem such a problem in differences being accidents of history. Whereas to account for the global organisation of the world as "just an accident" would be more of a stretch.

Of course many worlds would then be invoked to make the observed organisation of our world just a lucky nominalistic accident. In Humean fashion, there is no reason why the way things are couldn't fall apart in the next second or two. The laws of,physics might suddenly cease to obtain, for all we know.

However then it seems reasonable (to a holist) that the organisation of the world is self-perpetuating - a state of regularised habit - and so we would be back to saying extreme realism is the more plausible. Difference could be just accidents - matters of global indifference. The world would have enough sameness to actually be a world, and why should it care beyond having achieved that?

Even many worlds is still a hypothesis that presumes a substantial sameness in the overall shape and organisation of things. So in fact it only supposes localised differences - of the kind that doesn't really matter materially.

Thus extreme realism wins. Extreme nominalism always falls apart because of a lack of metaphysical glue. :)

The extreme nominalist would probably have a difficult, maybe impossible, task of explaining qualitative similarity. The extreme realist (or universalist) would have to explain why it seems as if there are particulars, and explain how our language talks about particulars without there actually being any. The extreme universalist, it seems to me, has it easier, as particulars could potentially be argued to be just loose bundles of properties held together by symbiotic powers and forces. To be crude.

In the end, it's all about taking what we are commonly acquainted with - particulars and universals, and trying to eliminate one of them by reducing it to the other. When in reality it seems as though both exist in co-operation. Is one "prior" to the other? That is to say, is one more ontologically simpler than the other? I'd say they're equally simpler and co-dependent; it makes no sense to talk about particulars without properties, and no sense to talk about properties without any instatiation relation between the property and the particular. Thus we arrive at a substratum, or substance, view.

The 'extreme nominalist' would have terrible trouble making a general argument.

for the sake of discussion, suppose that we must adopt one or the other. — aletheist

A nominalist who would adopt realism for the sake of discussion is an extreme relativist. His or her ontological commitment would merely be an adoption for the sake of discussion, not a belief about the nature of reality which, according to realists, exists independently of our discussions.

That's sort of the point. There are no general arguments. Each "general argument" actually picks out some specific truth. To have a group of people and say: "these are posters on The Philosophy Forum" is to talk about a specific truth about each individual.

General arguments are just incoherent. They tell a falsehood about what they are talking about: that a truth about a specific state of the world floats beyond it. (there are things like logic rules which don't talk about any state of the world, but they are always their own specific logical object.)

Thus we arrive at a substratum, or substance, view. — darthbarracuda

That is not one of the two options in this thought experiment. I was hoping to avoid debates about which (if either) is correct, and instead focus on how we would go about figuring that out, as well as the practical differences (if any) that would manifest between people who take each position.

A nominalist who would adopt realism for the sake of discussion is an extreme relativist. — jkop

I meant that each person within the thought experiment must choose either extreme nominalism or extreme realism, not that participants in the discussion have to pick a side.

General arguments are just incoherent. — TheWillowOfDarkness

Unfortunately for you, that is a general argument. It's really no different to saying 'generalisations are false', which is, of course, a generalisation.

Thinking and language require there to be categories, similarities, types, and so on. 'Extreme nominalism' (which doesn't actually exist anywhere outside the boundaries of discussion) would render those impossible.

The extreme realist (or universalist) would have to explain why it seems as if there are particulars, and explain how our language talks about particulars without there actually being any. The extreme universalist, it seems to me, has it easier, as particulars could potentially be argued to be just loose bundles of properties held together by symbiotic powers and forces. — darthbarracuda

Seems to me that it is very easy for the 'universalist' to give an account of particulars - each of those simply being an instantiation of some general idea. For instance, 'a table' being a flat surface, mounted on legs - you can then point to any table as being an instance of table-hood or a realisation of the idea of a table.

It's not a generalisation, but description. The denial of generalities does not take away similarities, types, categorisation and so on. It merely means they are something different than many people think: an expression of a particular rather than a constraint which defines them form the outside. Instead of reducing people to a single idea of what (supposedly )they must be, they are instead described in terms of who they are.

For the "universalist" it is indeed easy to give account of particulars because they just stop at their "universal," whatever they might it might be. They assume a particular similarity, types or categorisation must apply everywhere without checking what they are talking about. It's lazy thinking, an a attempt to reduce the world to one particular meaning, which can then be announced as a profound access to anyone in this world. Not only is it unnecessary (tablehood and a table are perfectly known and expressed in the particular), but it's actively ignorant because the reduction of the particular cases one to miss instances which are different or extent beyond that similarity, types or categorisation.