So just to diagram out this re, what kind of, the framework reproducible research I think about. Typically you have some published article that you view you know, in a journal, and that's all you get. You get everything that's in the article. You probably have some exposure to what their scientific question was, and these people went into into nature and it was all, and then there's nothing, right? And so, you have a, you know, you have a author whose kind of starting from here and went that way, and then the reader kind of starts to move there and wants to go this way. But for the most part, in most cases, the reader can start with the article, and then they're welcome to go back to nature, and just, you know, do it again, or replicate the study if they want. There's really not anything else that they can do, besides read the article and be impressed. So that's the express train to nature. So, but really, of course, there's all this other stuff, you know the data that I call the broad data or measured data. There's stuff that happens to that data, that, so a specific analysis can be done. There's all this code that is involved to do that, there's results, there's figures and tables, and this and that. So that all exist, but it's typically not available, and so the basic idea of reproducible research is to say, okay well lets all, lets just meet in the middle over here, and say let's make the stuff available to make the analytic data available, maybe, maybe make some code available, some of the preprocessing if it's important, things like that. So that's the kind of reproducibility compromise short of replicating the whole study. So, so I want to get a little into what problem this reproducibility solve. And I think one of the things that we get, from a reproducible study, is we get transparency of course, we have a better sense of what they've done. We get data, the data that they use, so that's available. If there's methods involved that are new, we get their methods. And we have this more, a kind of increased transfer of knowledge, because now we know exactly what they did, not what they meant to do. I, I, might be what we do, we don't get any sort of sense of whether the analysis was, was correctly done. And I think the main reason's because analysis can be very, be reproducible and still be wrong, and there are many examples of this in the literature where something was totally wrong but yet fully reproducible. So just because they don't, the people use the wrong methods, so they, you know, they kind of treated the data in the wrong way and whatnot. So, I think the fundamental question of, can we trust this analysis, is not really addressed by reproducibility. These things are all very important so I think that's fine. But we still have I think a little bit of a problem over here, so and one question you might legitimately ask is you know does requiring, if you required every study, every analysis to be reproducible, would that somehow deter people from doing that analysis. And I, I fundamentally think the answer is no. So, so some of the problems I have with reproducibility are, so the premise of reproducible research is that, you know, with all the data and all the code available to people, people can check each other. You can kind of validate someone else's analysis, and the whole system would kind of be self correcting in the long run. Alright. So one problem which I don't see here is that the long run sometimes is too long and then in terms of the context of the problems that you're dealing with I think reproducibility addresses what I call most, the kind of downstream aspects of scientific dissemination. Now, I'll be more specific about what I mean by that. Meaning that it kind of only happens post publication [INAUDIBLE]. And another key thing which is important in my area, maybe. I mean, particularly in my area, is that, [COUGH] the ideas of reproducibility kind of assume that everyone plays by the same rules, and everyone wants to achieve the same goals, which is definitely not true.
