Carl Heneghan

It's like we're on TV in a big advertising channel and we're not affected by what we buy.

We're just being marketed at too much.

Separating out and going to events where you can truly think about the issues and you don't feel you're commercially sponsored is, I think, an important issue.

Anybody who wants a piece of work out there right now, what we need to work out is how much money is being spent on educational activities that's funded by everybody, governments, industry and all sorts of people together and then say how much would it cost if we replaced that with some fund.

And I think that would be quite an interesting activity in 2020 to try and solve.

I'm going to go quiet, everybody.

One of the developments in the last decade has been big data, artificial intelligence and our use of databases in ways that mystify me on a daily basis.

I think you can tell absolutely nothing from this paper because it's big data gone mad.

Look, it's telling you about the relationship of the first and last offer when they're female compared to the first and last offer when they're male.

So that means the middle of the paper could have been filled with men or women.

And to be honest with you, the last time I published the paper, every single person contributed to the draft.

So the idea here is it suggests, if this is true, that actually all the people in the middle shouldn't be authors.

Well, we sometimes, the second author can be just as important.

Yeah, so this is one of these papers where they've just lumped together a load of studies and you've got precision over accuracy.

It would be much better.

What it means is you can create numbers that actually have that are statistically significant because actually the more you sample, the variation gets smaller and smaller.

So that difference is significant, but actually might not mean anything because it's some way from the actual target of what you want to understand.

So it could be much better to go in depth to about 200 papers and really go in depth and analyse what the issues are and look at that in depth.

But this is something you can run on a computer really quite simply with a big data technique.

And I'd much prefer somebody to go to me and say, look, are we going to do in-depth interviews?