This blog post is a quick run through argumentation, why I think it matters and what I’m trying to do as part of my PhD. Hopefully, I can sned people a link to this instead of doing the same spiel everytime someone asks me what I do.

Argumentation is a field of study that fundamentally concerns itself with processes of reasoning. It seeks to answer a deeply epistomological question: given what I (and others around me) know, what conclusions can I (or we) draw?

In the world of classical logic, barring the bickering amongst logicians of which theory is the best, this question has clear cut answers. But, classical logic makes a whole load of assumptions about knowledge, but primarily, most logics require knowledge to be consistent. Any inconsistency, like “I think that I am happy, but I also feel sad” causes the whole world to implode in on itself and truth collapses under its wreckage.

I can say for a fact there are days where I am simultaneously happy and sad, where actions are simulatenously moral and immoral and a hotdog is both a sandwhich and not. The world is full of inconsistencies, yet here you are reading my little blog post (probably because I’m standing behind you nodding my head and asking you what you think).

Postmodernists be damned, most of us also recognise that some truths about the world still exist. So if the world is inconsistent, if we all disagree all the time, how do we know what’s true?

At some point, (even before the invention of classical logic) somebody made the miraculous discovery that we all have reasons for the beliefs we hold. And we really like to make others agree with us by yelling our reasons at them and covering our ears until they give in to the barrage of noise and say “sure, whatever”.

Now, this system is great, especially when its me doing the talking. Sometimes, unfortunately, we have to concede and hear people out.

In this case, how do we know who and what to believe? who’s giving better reasons? why should I be convinced that Eppstein didn’t kill himself?

This is what argumentation theorists get paid to argue about. One of the main ways of doing this is by studying the common patterns of argument that we come across when we look at data and identifying the ways in which these arguments could be called into question. This kind of critical questioning has two challenges, in the first case we need to know which set of critical questions apply to the argument at hand and also, which question(s) is(are) the most relevant to challenge what has been said.

I imagine that applying a systematic method of critical analysis in every conversation can be tiring and unnecessary. People engage with insane quantities of information on a daily basis and are not equipped with the mass of content they have to engage with. This could partly explain why we live in a post-truth world. To actively engage with and analyse discourse is just too cognitively demanding, especially when people have other demands on their time that don’t involve thinking about argumentation.

I think this is where computational argumentation comes in. One way this can happen is to identify the arguments and present users with critical questions that can help them think more deeply about the content they are engaging with. This requires computers to be able to automatically identify the pattern of argument that’s being used, which is really really hard, even for people.

The main reason this is hard is that people don’t say everything they mean. We often assume common knowledge, or purposefully leave out questionable assumptions to avoid them getting attacked. Figuring out what information has been left implicit is basically a form of telepathy since we can never know what’s going on inside the other person’s head assuming they are even aware of the assumptions they are making in their argument.

This is where my project comes in. I’m trying to figure out how to build systems that can identify the most reasonable assumptions that are needed to make an argument work. Basically, I’m doing a PhD in mind-reading.