Not sure if Afforess has looked at this yet, but I recently had a really bad example in a C2C game. Essentially it allowed me to go from a difficult (but interesting) situation to an easy game very fast. I was the power chart leader of the 4 civs I knew, but one of them was significantly ahead technologically (like at least 100 turns of research worth), and from 'Most Advanced Civ' pronouncements I knew he was nowhere close to being the tech leader, so things looked like they were going to be an up-hill struggle longer term. Anyway, he suddenly decided he was prepared to trade most of his techs (I'll go into why later, but I don't think there is [much of] an issue there as it turns out), and I was able to offer him a resource he lacked, for which he gave me 4-5 civs. I was able to end the agreement every ten turns and extract a similar batch of techs each time until I was pretty much caught up after about 30 turns. This is obviously rather exploitative
2) The real problem is that BTS originally never allowed per-turn trade items to be traded for permanent items (e.g. - resources-for-peace-treaty-time for techs), and when this was introduced the scaling of the valuations between the two categories was a bit arbitrary. In particular the code is loosely intended to normalize everything to a gold value, so for gold-per-turn for example it just multiples by the peace treaty length. However the valuation of resources uses the same routine that the city AI uses for assessing resource importance which has no such normalization concept (as its just used to compare against similar values really). In particular it scales in number of buildings and units that have some sort of dependency on the resource, regardless of their OTHER pre-reqs, so as buildings and unit types are added (relative to BTS AND adds a lot, and C2C a lot more still) the resulting values tend to inflate.