What makes a conversational interface

Note: This was initially a comment on this great piece by Matthieu Varagnat on Conversational interfaces beyond the hype. It triggered a good discussion, so I thought I would flesh it out here.

We’ve definitely reached the Peak of the hype cycle on chatbots and conversational interfaces. I mean, we are past the point where every third startup pitch involves an AI, a bot, a conversational interface, or a combination of said features.

However, I feel like we’re lacking a thorough reflection about what actually makes a conversational interface.

I disagree with the fact that it has to be text-based, for instance. Or that it has to necessarily involve several people or bots.

For me, the essence of the conversational interface is to present an exhaustive chronological representation of past actions.

It can be text-based, but it doesn’t have to be. It can be multi-user, but it doesn’t have to be (I have a Messages thread with myself, for example. Granted, I even have a Slack team for myself).

But the defining point for me is to have access to a perfect transcript of past exchanges and actions: the information is complete and the chronology is linear.

I think the current dominance of the messaging interface is a reaction to how complex feed interfaces have become. Feed interfaces (think Facebook) have nesting (posts and comments are on a different level and have different properties), are partial (they are algorithmically filtered) and non-linear (stuff pops to the top, according to non-obvious, non-deterministic properties).

Conversational interfaces just make my brain way happier.

What do you think?