When I was at that gaming conference, populated as it was by English academics and the like, I was bothered by the overuse of analogies. Sure, I liked the analogy between gamer shame and queer shame, for what it was worth, but that was one hit among several misses. I felt much less enlightened by the analogy between video games and candy (and subsequently between candy and sex).
Indeed, one of my initial reactions was, "Wait, can analogies even be used as arguments ever?" You can introduce and illustrate new ideas through analogies, but can you really demonstrate anything with an analogy? If you observe that X and Y are similar to each other in some ways, this does not demonstrate that there are further similarities. When two things are analogous to each other, you can never say where the analogy ends and the disanalogy begins--not without observing directly.
On the other hand, it would be wrong to categorically dismiss analogies as arguments in all circumstances. I'm sure there are examples of proper arguments by analogy out there, even if I can't think of them in that moment when I'm blinded by English professors. It would be helpful to consider a few recent examples where I used analogies on this blog, and critically examine how I used them.
In Oppression Olympics: A balanced perspective, I make an analogy between the way that people get excited about that one trading card they found in a booster pack, and the way that people get excited about that one argument that they were able to think up on their own.
This analogy fails as an argument. It does not demonstrate that people have a tendency to get attached to arguments, the way they do to trading cards. I was merely asserting that this is the way things are, and hoped that readers would agree. If I wanted to present a real argument, I would refer to psychological research.
In Negative may be better than the alternative I made an analogy between the "negative" labels atheism and asexuality. I said that people have come up with "positive" alternatives to atheism, and that these alternatives have had both costs and benefits. I then argued that the costs and benefits would also apply to asexuality.
I think this comes closer to a valid argument from analogy. The key point is that there are underlying patterns in the way we interact with identity labels. Therefore we can predict a certain amount of similarity between them. This is by no means a perfect argument, but then any argument about social trends is going to be messy.
In Why video games are so flammable, I have a brief simplistic discussion the economics of video game consoles. I model it as monopolistic competition with a strong economy of scale effect.
In a way, every argument based on a model is an argument from analogy, because I'm analogizing it to that model. Even if I'm doing something as simple as adding up money, I'm making an argument by analogy because I'm analogizing money to the abstract mathematical concept of numbers. This is a valid argument, because economic models are built on a certain number of premises, and we know those premises are approximately correct.
Based on these few examples, here I will draw some conclusions. Analogies are often not used as arguments at all, but rather as tools to illustrate concepts. However, there are cases where analogies can be used as arguments. Arguments from analogy are at their best when they most resemble arguments from models. If you want to argue that two things are similar, you can't just observe a few similarities and hope that other similarities follow. Rather, you argue that the underlying patterns or laws are similar, and therefore the consequences of these laws should be similar.