I think the difference between the u and o endings for objects are whether the object is one of a choice of objects, or whether it is not a choice of objects and just an inquiry.

My take on it

-iu asks the listener/reader to make a mental selection either in their interpretation of a statement, or in their answer, choosing from a list (the list is already established by context, e.g. outside a pet store "oh look at those puppies, which would you like to take home?!" "oh that one there, with the big cutey eyes, naw look at that tail waggle!"

-io does not require the reader/listener to figure out its meaning by selecting a choice out of a list of possibilities - it's more ambiguous (one could say that the list of possibilities the reader can choose the meaning from is everything in existence, or, there simply is no assumed list).

Tio is a demonstrative pronoun; [/i]tiu[/i] is a demonstrative adjective. Because tio is a pronoun, it functions like other pronouns, i.e., as a substitute for some noun that is understood in context. Since tio is not generally used to refer to persons, the pronoun that it most resembles functionally is "it".

Because tiu is an adjective, it should be thought of as always "bound" to a noun, even when the noun is omitted and it's used alone. Expressions like tiu hundo and tiu domo are the models. When it's used without a noun, it's understood that the implied noun is either "individual" as in any particular one out of a group, or "person", where the reference group may be humanity itself.

Tio estas mia domo.
Tiu estas mia domo.

The first sentence conveys "That is my house", in the sense of "that thing over there," where perhaps the viewing conditions make it unclear that what is being referred to is a house at all. The second sentence conveys "That is my house" in the sense of "that one", out of a group.

Tiu estas mia frato.

You wouldn't use tio in such a case, for the same reason you wouldn't normally use "it". And yes, I'm ignoring the sort of case in English where the phone rings and I pick up and say to you "It's your brother."

tommjames:It's interesting to note the contrast between written works and colloquial speech; where Tekstaro has "kiu" as the most common correlative with 17.72% and "tio" in third place with 9.41%, the logs in the chatroom at Freenode actually have "tio" as almost twice as common as "kiu" and the 12th most common word out of all words spoken!

Could that be because the chats at Freenode generally consist of short and simple sentences?

Chainy:Could that be because the chats at Freenode generally consist of short and simple sentences?

Well it could just be that currently the logs only extend back by about 8 months and so the stats are inaccurate. But I don't think it's that, rather that conversational speech makes greater use of certain words due to stock phrases and common themes of discussion. Some words which are obviously going to appear more often are saluton, kiel, vi, farti, mi, bone, dankon, ĉu, jes, ne, etc.

How "tio" fits in I'm not sure. Actually I was thinking of extending the channel bot to do all kinds of text analysis where you can get stats on common phrases, and see which phrases certain words appear in most frequently. I may get round to that sometime soon.

Perhaps "tio"s frequency is because in casual conversation, you can't think of specifics as easily as in a planned text, so you often rely on "what's" and "that"s and "this" etc. Especially in spoken conversation where you're on the spot and can rely on primitive body language to get through the day without having to think about the language you're using