 Good morning Welcome to day two of meta science 2023 Thank you everyone yesterday for your attendance your conversation the discussion all the presentations were excellent and the discussions were lively And so it's just wonderful to see the engagement That we've been having so far At this meeting and we have lots of great things to cover today As yesterday we have concurrent sessions the next one will start at 9 30 In the same room as before conference room 120 And the 9 31 will be using natural language processing systems in research Challenges and opportunities. So if you're interested in that as a dynamic discussion, please attend And for this morning session Unfortunately as many of you may have heard samine vizier who is going to give our opening keynote It has coded so she is not here and she is not Going to deliver that I will put up a link to her blog where she has posted The talk so that you can see what we have missed regrettably and Of course, we were desperate. So I'm very Apologetic that our replacement speaker will be Brian no sec today So as the moderator of this session, it's my task to provide the introduction Brian grew up in Davis, California in the shadow of his younger brother Davis high Kevin the younger brother was all-american water polo player and captain of the basketball team He went on to play at UC Davis For basketball. I had a very good career there and then became a Coach at UC Davis Friend has developed in his career and is now associate head coach at UC Davis men's basketball team and This meant that he was in the newspaper all the freaking time And it's so much so that when he got married There was a full-page story about his wedding in the paper like he's a freaking celebrity and I'm convinced that if Brian ever gets an award that should earn coverage in the local paper It will have the headline coach. No sex brother gets a word Brian is totally fine with all that he's not bitter Perfectly great thing. He's proud of his brother. Yada. Yada, whatever So please help me welcome to the stage coach. No sex brother Thank you Brian for the worst introduction ever Samine's talk which would have been awesome is available at the link above So please go do check that out if you just Google Sometimes I'm wrong blog it will come up if you don't get that down or just snap a picture What I'd like to Talk about today is the general interest in how is it that we think about making research trustworthy and I want to review some examples of how the public responds to different scenarios very simplistic scenarios of researcher behaviors and how that impacts their perceived trustworthiness and Then talk about how that may have some object lessons for how it is we think about trustworthiness more generally And perhaps in in a very systems sort of way So one way we might think about how researchers become Trusted in public is if they're confident in the claims that they make they provide certainty of those and of course That the claims that they make are correct and the base argument that I want to Start with is that that's the wrong way for us to think about how researchers gain trustworthiness in the community for the validity the applicability the relevance of scientific claims for public consumption Instead, I think the model that we can aspire to for earning public trust in research Is to cultivate a community where humility Calibration of claims representing the uncertainty appropriately and Having as the primary objective true seeking rather than true certainty is the way that researchers become and will be trusted So the context that I want to provide for this is a survey that Charlie Ebersole Jordan axed and I conducted on a broad sample of US adults several years ago Where we gave them very simple scenarios and ask them how much ability do you think this researcher has? How ethical do you think they are? How likely is it that you think their claim is true? And then we just altered the scenarios in simple ways to see how the respondents would react to Those three variables given what the researcher did or what happened next? So the baseline question that we had is at the bottom of this here Researcher acts found an interesting result and published it So the baseline of how much do we think this person is a good researcher able? How much do we think they are ethical? How likely is it that they're finding is true is always against this baseline of I found something interesting and I published it and If the scenario suggests that there's an it perceived increase in perceived ability ethics and truth It goes to the right of and then if if the scenario suggests that those are the Researcher is less able as a consequence of other things that happen than it goes to the left So simple first follow-up information is I published a result interesting result and Stewart Succeeded in replicating it some other research you did what happens the perception of my ability It's always in reference to my ability as the original researcher I'm perceived as more able the a that's covered by the e I'm the perceived as more ethical compared to that baseline and it's more likely that the finding is true Likewise if Stewart fails to replicate my finding what happens well I'm the finding is perceived as less true and I'm perceived as less able and slightly less ethical Just as a consequence of him failing so in these two very simplistic scenarios We might have the general conclusion that really our perception our reputation Understanding of our trustworthiness is tied to whether we produce truthful claims or not But I think if we expand this very quickly we see that these are separable So let's add complication to those scenarios Stewart fails to replicate my finding and then I say Stewart's methodology was wrong and his result isn't valid What happens to the perception of me in that scenario? Well, I don't affect the perception of the truth value of the claim It's still people think about it the same way if they just knew that Stewart failed to replicate my finding They now also think that I am less ethical and less able than that baseline of having said nothing at all But I could have said oh man Stewart's right So I agree with his methodology and I say oh, maybe I was my initial result was wrong Here we see a substantial separation Research the public thinks that finding is now even less true because both of us agree with that But my ability and my perceived ethics Are not just back to baseline of I provided a result But in fact they exceed the baseline of all's you knew is that I had found a result and published it So even though the new evidence suggests I am wrong. I am perceived as even more able as a consequence Or I could have stored might have failed to replicate. I start a new study say I want to figure out Why Stewart got one thing and I got another thing in this case my perceived the perceptions of my ability and ethics are the highest among these and The truth value perception violates Bayesian principles all as they know is that there's now new evidence of failed replication but because of my response the perception is that oh, no I think I've even a little bit more confidence in the truth value of that claim despite the failed replication Because of how I am responding to that evidence and pursuing more information Doesn't require that Stewart fails to replicate it can also be just my behavior So I publish a failed replication challenging my original result. I published one now I published a second but now it's not true truth values just as responsive as if Stewart is the one that did it But my perception of my ability and ethics again go the other direction because I'm just trying to get to the truth But if I said oh, I failed to replicate it, but I thought was bad study. So I'm gonna put in the file drawer The negative consequences are just as the same or even more substantial as if Stewart had done that But and I just criticized it and then finally if I just didn't even bother following up I published something and then went on to the next problem instead of trying to follow it up This lowers confidence in the truth value of that claim and lowers perception of my ability in ethics So in total especially this final one it represents that public understands that there is Substantial uncertainty in the claims that we make and the initial studies that we do and that following up is actually a normal Ordinary part of the process. Let's figure out try to root out error and understand where and when and how These things occur and under what conditions they will emerge What I want to call particular attention to are these scenarios where The correctness of the claim and the percept the reputation of the researcher are fundamentally separated from each other That it is not about being right It's about pursuing Correctness it's about how it is that we respond to evidence that challenges our initial ideas That is the driver of the perceptions of ability and ethics and so the core local observation is that in these types of scenarios at least the public responds to a Orientation of getting it right over an orientation of being right for what makes a trustworthy researcher And I think there's a more general point to make which is the trustworthiness of research is much more about the process of The research rather than about the outcome Outcomes in science will always be the basis of distrust because science sometimes presents us with things We don't want to see that they challenge our ideologies our perspectives our financial interests There's lots of reasons that I hate some scientific findings because of personal interest I have in those findings and so if we aspire to a research community where the public just accepts our findings Because we're saying that no we know this we can be certain about this We'll never meet that standard and we shouldn't meet that standard the fact that we have pushback and skepticism and responses entanglements all of that is the productive conversation of science and There will always be some reaction to the truth when it's different than the realities that we hope for or that we want But that's of course not the right battle to fight is to have the public just accept our claims uncritically The approach to trustworthiness that we can pursue and embrace is one of being resilient to that challenge Trustworthy research is the is the work that prepares the research so that when it is criticized Valid criticisms can come through because you can see how the work was done and identify where those weaknesses are But the evidence and the availability of that evidence then is resilient to those attacks of I don't like This finding and I don't like you for producing this finding because those things will occur And so if we can have a substantial evidence base that provides a bulk bulwark against that That's how science maintains its trustworthiness and the reason that people get so upset about scientific findings that are unwelcome Is because science has been so successful in its history in developing understanding So it is a threat to the ideas and positions and wants that we have now But it can survive that and thrive in the face of it, but the Previous scenarios might suggest that this is all just on the researcher Themselves to figure out how it is to make sure that their work is trustworthy But of course we are embedded in the broader systems of how science operates and rewards researchers for its work And so how we achieve trustworthiness is not just on the individual researcher But also on how it is we create and manage that system and there are serious threats in how it is our system is managed now That can undermine the entire Trustworthiness of the enterprise and so I want to give a case example of that to prompt some discussion in the context of paper mills and predatory journals paper mills if you're not familiar with this term are Authorship for hire I have a paper. I'm gonna give you third author if you just pay me 500 bucks and there's lots of these and Someone that's been looking at this Thinks that there are hundreds of thousands of papers in the literature that are a consequence of paper mills often fake papers That are just selling authorship and sometimes potentially even real research that has sold authorship and collusion rings with editors and journals for selling this Right, we just now get the author to pay. We'll publish it in this ridiculous journal Predatory journals are a more general concern, which is no peer reviewer just willing pay for play if you're willing to pay We'll publish your work So there's lots of interest and work on how is it that we can deal with these types of problems And a lot of the solutions are looking at ways that make it increasingly Inconvenient for doing misconduct making up papers inventing data But have a predatory publishing etc. And that these solutions make it easier to detect misconduct So for example, if the paper is open access, maybe a little bit easier to figure out When when these things occur if the if they have disclosures and reporting standards in the paper Maybe that makes a little bit more inconvenient for people to fake these things if the peer reviews have to be public And there's these collusion rings. Well, that might make it a little more inconvenient to hide That this is in fact a collusion ring if they have to share the data and measures protocols, etc the more that's visible the more work that the misconduct has to Overcome in order to be convincing that this looks like real research even though it isn't The key here of course is that this is increasingly inconvenient and it's helping to detect So those are good things for trying to address this corrosive problem Especially if we're talking about the hundreds of thousands and potentially into the millions of papers That are now a basis for not knowing how to trust the entire literature But those detection methods ultimately only treat the symptoms of the problem and the underlying cause Why is it that predatory journals exist? Why is it that paper mills are successful? Is the reward system itself and if we don't Try to address that long underlying cause that it'll just be a continuous arms race between those that are commercializing fraudulent work and Those that are trying to uphold the integrity of the research literature The basic model that we might think about for why is it these things these Practices predatory journals and otherwise are thriving is that it starts with a researcher needs to publish For their for their job for their advancement for having a career And if a researcher has Resources to do that research has time to do that research doesn't have an overwhelming teaching load Has training to be able to conduct that research is part of a scholarly community in which those ideas get generated and Developed and has the support in their institutional contexts higher bees space to do it follow those things are available then Most researchers get into research because they care about doing research and so they'll do rigorous research And then they will try to publish it But in many contexts one or more of those things aren't available And yet there are demands on the researchers to still publish still have to produce things I don't have any resources. I have a 5-5 teaching load. There's there's Where do I do the research? What are you talking about? How do I how am I gonna do this? So if I don't have it though some of those things then predatory journals paper mills and these other sorts of solutions Actually fill a need for me as the researcher It gives me a way to meet this demand that is on me in order to have a job So it actually is serving my interest Even if I know transparently know that this is all Fake because if it hits the goals that I have for what are imposed on me through my institution or otherwise Then I kind of am stuck and it's not surprising that I might go that direction And that's the reward system there is a consequence of multiple agents universities publishers societies funders all Influencing what researchers do and those researchers have a mix of motivations. I need money and I need a job I'm interested in having a good reputation. I want to make a contribution to science And I want to feel like I belong to look this community But first putting those motivations against one another is a challenging thing if I will have no money and no job Then how much can I prioritize? Making a real contribution and So we could hope for we could demand that people change their motivations You shouldn't it's to be ethical you should not care about having a job. You should do the right thing And that's a big ask We know that these base needs are fundamental needs So that's not the right tack to take the right tack to take is to change how researchers earn those rewards and to adjust the Stakeholders that are driving those rewards imposing them on researchers themselves But the problem is even deeper because those Institutional agents also have motivations motivations to have money motivations for reputation motivations for contribution motivations for belonging So institutions that are in Nations or environments or other places where they feel like they need to increase their status may impose Those expectations on researchers Because they have to address the motivations that come from their own context right the government policies that say We need to be better in research in our nation. So dammit produce more research University says well, we don't have any money So we'll just make a requirement for researchers produce more research Or if they're worried about their ranking systems and the ranking systems actually credit how many publications you have It's no surprise that institutions who are competing for students and competing for money will be responsive to that So if we don't tackle those fundamental problems We will always have this challenge of researcher misconduct and misbehavior So it is great for in the now To make to pursue detection make these types of misconduct more inconvenient try to expose them and address them But we need to continuously be working on the medium and long term of Changing what is actually required to achieve publication for example the model of registered reports is less convenient for Fraudulent behavior because you have to prepare and have reviewed in advance what you're going to do and then you have to do it And then you have to publish what you did And in the long term We have to be working on those institutional agents because if we don't change them none of it will change So I want to end noting that the reason that science ends up being trustworthy is because of its constant self skepticism its willingness to confront itself and its evidence for how it is we can become Trusting of that evidence what survives that inquiry that we are all part of and Elevating that is helps us Elevating the argument and the disagreement and the challenge is what makes science lively and Also makes it so that we feel like okay. What is able to get through that system is stuff that we can really count on So thanks for tolerating the replacement and thanks. I'm happy to have any questions Dan, please first question. This is exciting So the plus biology paper I think that's the one that you used to frame this talk is about asking people how they update their beliefs about Truthiness and so on our truthfulness In the abstract of that paper you describe the sample population which is undergraduates and I think other researchers So I was wondering what can you tell us about work that's followed up on this looking at other populations? for instance the general public and Is there any results around the kinds of people that update in one way versus another? Great. Thanks. It is cruel to ask about the actual abstract because that means you read the paper And I can't just BS my way through the answering but that's great great question the data I showed is a general public sample the Sampling group claims that it is Representative and I don't believe that it is based on the demographic data. We have all these sampling firms claim that they pull representative samples And they do not that's a different talk So it is a general sample the others other studies in that paper that are focused on undergraduates and researchers and that's Evaluating Doesn't matter. It's evaluating other parts of research credibility But the point still remains is how much do these extend it and we don't know we've not followed up on this. Uh-oh that means For our reputations. Okay. We will follow up on this because that's a way for us to be more credible in our work but the the core is that they Across this pretty diverse sample. It showed the pattern was very consistent Wherever people's experience with science and otherwise, but it'd be really fun to try to pursue much narrower domains where people might have different Points of view on science in the first place and see how they are responsive to this. Do you have any hypotheses about that? I I don't but thank you for your humble answer and I'll update my priors Please Systems level discussion critique was was really really on point But I see here at the ends we didn't throughout the talk. You're sort of conflating science with the contemporary Academic science. It wasn't always this way and it doesn't always have to be this way Certainly, we're seeing the increasing politicization of science and specifically academia in certain American states and throughout the world And so I take the point that we can't actually be closed and ask people to trust us, especially given the internet But I wonder in the experiment you run about where we should choose to be open so the stimuli you were showing we're discussing individual papers being replicated or not and certainly we've seen a lot more of Media reports about individual papers being replicated or not in the past decade It doesn't seem at a macro level that trust in science in that decade has gone up, right? So my question is where should we choose to be open and and at what point do we need to have the public involved in what we're doing? Versus at certain points what could be done behind the scenes? Yeah, that's a great question and my default answer given my organization and role is that it all should be open all of the time But maybe not so I So this is point of view speculation is that I think we benefit from radical openness if our aim is trustworthiness and the reason is is that people care when it's relevant to them to care and That's a good thing in the sense that if you make the entire process open Well, when am I going to wait in to actually interrogate it that evidence and respond to it and be reactive to it or whatever else? It'll be when it matters to me and so just within the research community itself I am highly attentive to things that are in my research area And I might even pay attention to it at the onset of some of this work But I stuff this just on the side like a motion research I don't even pay attention to it until there's a review because I don't do I'm not an emotion researcher But I know that I need to kind of understand what's going on in the motion field for the things that I think about And I think the same plays out in the public for broader issues Where I think I'm wrong about that is on highly controversial issues. So research on abortion gun violence, you know the things that really are the polarizing topics When that research is open right from the start It creates more avenues for bad actors to potentially engage it And I I want to still think that openness there is the best way for science to go But there are obvious risks and even like health and harm types of risks that people in highly controversial areas will confront So I don't know what the right answer is there Do you? That's I think that's great You know, I think bad actors is it's political actors and we are political actors And I think we can't pretend to not be political actors. And so yeah, thank you tough question. Yeah, thank you appreciate it. Yeah, please Martin Bush University of Melbourne Brian, you've spoken as eloquently as you always do about the need to shift cultures and incentives and supports But it seems that at the end of the day We're still talking about the unit of credit for science is the individual publication. Is there Do you see no future for academia where all of these other behaviors which support? credible science Genuinely part of a reward system contributing to collaborations in non-research ways being a good mentor providing service to societies and institutions I'm optimistic that it can be diversified To be better include those things and I think they do include at least in my own Experiences of very direct evaluations hiring for new faculty in the Department of University University, Virginia our tenure process what I observe is attention to all of the things and Nevertheless that primary currency is the publication in a research institutional context And I don't think we can get rid of it as a key currency easily Because it is a useful mechanism of communication It is a useful way of summarizing there's different models of what a publication is right micro publishing We can think about different components and units. We can diversify what a publication is but I think the products of research Will be hard to not have be a critical incentive For researchers because that is what we're doing. We're producing some kind of products so I start with that assumption of If that's at least in the terms of that we can foresee Will be it will be part of the a key part of the incentive system. How do we address that and Then if there are other models that emerge where you don't even need that we can all just edit Wikipedia together Awesome, let's try try those and see what can scale up and how it is that credit could be allocated in an effective way So thanks for that. These are all very very hard questions that I wish you hadn't asked because I don't know how So, please Thanks for a great talk Ted hodap from the Gordon and Betty Moore Foundation I'm just gonna ask you about the full-court press. That's for your brother So you're talking about the the pressures On what I would kind of say this side of the the argument What about the other side of the argument? So how does the pressure that we impose within the system? Help to make the science better and so can you speak about that a little bit? Yeah, great Yeah, because the the strength of incentives is also a force for good Right and I almost always frame this in terms of the dysfunctional incentives because We recognize those and we complain about those so much. How do we fix it? But the whole point is to align the incentives not to get rid of them But to align the incentives with the values that we have for scholarly work How do we make it in a researcher's interest to be more rigorous and transparent and reward them for that? Because rewards matter we might idealize the concept of oh if we just eliminate reward systems Researchers will just do it through genuine intrinsic interest and we'll all learn but that's obviously not the case We we need those reward systems as part of natural human behavior And so the key I think is how is it to think about how is it that each of our agents of Driving incentives can shape those in a way that really promotes Productive collaboration productive competition productive engagement on the ideas that we have so thanks for calling that out I appreciate it. Thank you. It's time for one more question I'm in a hatch from HMI I'm trying to square sort of a couple of slides so the The systems level diagram you had with at the top there are Pressures on institutions such as university rankings and then the long-term strategy Of course is changing the incentive system and I'm wondering how rankings fit into that strategy Yeah, so the is a great question and I would love for there to be some Summit meetings between those that create and manage the rankings and those that have thought about these issues of how the rankings are Presently dysfunctional, but maybe could be more functional to see if there is any Any way to close the gap for making Transforming those the challenge of course is that the people that are operating ranking systems us news and in the US and the more global ones They get a lot of benefit from having those rankings. It is really in their interest to have them And so if we're going to make any progress on them, I don't think we can say you need to get rid of your ranking systems Or that we're gonna stop caring about them What we need is for them to reform those rankings at least in the near term So that they're actually again Incentives aligned. What is it that we want universities to be and can we provide create indicators and metrics so that Universities can strive to be that rather than hit the benchmarks that are presently driving those rankings the I would love to think that we could just remove care about those but universities also have to be responsive to it because The consumer is responsive to it. The student pays attention to those things So I don't think we'll get rid of them. I think we have to reform them, but can we I don't know So thank you. Okay. We are past time. So thanks very much for