 We will have the honor to have Richard Gu to finish the day with the talk about historical consistency, causal relevance, and historical token causal description, a causal model modeling account. Please Richard. Thank you, but authorizing the abstract actually was all a paper 80 percent. It's a new one. So it's my pleasure to be here to present my ideas and thank you very much for allowing this to happen. But Alex, I won't talk about metaphysics. And for some reason I don't want to say that I'm going to talk about methodology as well. So I'm just going to present some very basic ideas. And I actually construct seven formal conditions. Some of them are very worthy. So I will use either drugs or a case to explain the formal conditions. And for this historical causal conception, it's all about human history. I only talk about human history. So some basic ideas. So this is the case. And this is the only case. So had Champlain implemented an entire piece of the policy, he did that back down from his data. So this is a historical causal description inquiry. Whether or not this is suitable. So normally we use two variables, a cause variable, a fact variable, and try to see whether or not we can build up some sort of equation to explain. But for historical studies, it's a little bit different. They ask for the so-called historical consistencies. They'd like to know whether or not this kind of fact was likely to occur in actual history, in actual course of history. They are not doing metaphysics. They are doing fact-based science. It's not metaphysics. So the thing is that how to explain historical consistencies, how to formulate the idea of likely to occur in the actual course of history. So this, they want to capture the peculiarity of historical possession. And history is famous with manipulability. You can't manipulate it. So how can I use a causal model in a country to do that? That's oxymoron. But the thing is that the proposal is using a historical thought experiment. That is, depending on causal generalizations or theories of relevant disciplines and historical facts as evidence to create a ground to justify a thoughtful intervention. So that's the idea. And the method, method, is counterfactual backtracking analysis. Now what is counterfactual backtracking analysis? It's backtrack from the charged counterfact, historical counterfact to its precondition. So apparently, those relevant disciplines, the causal generalization theory are going to assist to establish what is the precondition of which. So that's the role they play. And continue the backtracking process until we find a historical fact to anchor the backtracking process. So that's the basic picture. But there's a lot of things you need to explain what's going on here. So I set up a probability distribution requirement. That is, I asked for the probability distribution between historical counterfact and its factor counterpart to be approximately equal or even higher. So that's the idea. Now you may say that, well, it's too idea, right? But my job is to formalize those causal conception, those causal conception. And give this distribution, I can do that. And what is the suitable, reasonable probability distribution threshold? Actually, it depends on historians' expertise, not depends on philosophers, right? So I just set up an ideal condition and based on that, to explain how to perform historical causal conception based on causal model. So this is the case. So this is apparently a counterfact because Chaplin only implemented a piece of policy. So with backtracking precondition, and they said they need a strong national defense force. So you can implement some unyielding diplomacy. So they think that, okay, in this case, the present case, this is the UK's official dominant after the First World War. And it is still a counterfact. So they want to keep backtracking and backtrack to its precondition. And they think, okay, a hawkish cabinet might think, not mine, must think that the strong national defense force is essential. But still, that's a counterfact because Chaplin led a dovish cabinet. So they backtrack to the precondition of this hawkish cabinet. That is, they are hawkish cabinet members occupying substantial positions, and even becoming the prime minister. And this is actually a historical fact. There are three. There's Eden, Cooper, and Churchill. So there are three. So this is historical fact that we can use to end the backtracking process. But the thing is why and how. So this is the graph so far. This is the graph so far. That is hawkish cabinet, sufficient rearmament, a peaceful anti-epicement policy, and the success or non-success of sedentary defense. So that's the graph so far. And this is the variable setup. And I don't need to go into the detail. And this is the equation set. This equation set is actually saying this. Respectively, hawkish dovish cabinet causally connected, insufficient, insufficient rearmament, causally connected to unyielding, yielding diplomacy, causally connected to heat loss back and down or not back and down. So that's the equation, one of the equations set. Apparently, the causal profile of the variable actually is not fully explored yet because we can't see FCM plays any role yet, that is hawkish cabinet members. So my idea is, I assume that this backtracking is a successful backtracking. And I want to, based on the causal modeling, to formalize the correctness condition of a successful backtracking based on which I can explain the advocacy of historical existing inscription. And on top of that, I can explain the advocacy of historical causal inscription. So that's the basic idea. So as far as I can give those correct conditions, then I can explain that each historical counter fact, their consistency. And therefore, if two historical variables, one is the Alice immediate incentive variable, and both one value of them represent counter fact, and are both historically consistent, then the causal inscription is there. So that's the rough idea. So the most important job is just to establish those correctness conditions. That's what I said, the seven formal conditions. So now, seven formal conditions. And we can see that for each backtracking, they can, each backtracking, it consists of two parts. One part's backtracking to a historical fact, and another consists of backtracking to a historical counter fact. So we can talk about it. They have different formal structure. So first of all, a fact. So this is FCM. So that is hawkish cabinet members occupying pivotal position, blah, blah, blah. So that's the anchor in fact. But apparently, by itself, you cannot play a full causal to the hawkish cabinet, because the actual cabinet is dovage. So it can't play the full causal, otherwise it's equal hero inconsistent. But we have reason to say that at that historical juncture, we can assume that something can represent the exogenous variable that we can thoughtfully intervene in. So that's the idea. We can add FCM here, and this one is the exogenous variable. And this exogenous variable will represent the contingent factors related to the tenement of the hawkish or dovage cabinet. So it probably doesn't relate to the party election, some contingent factors. The second of all, the default position is for its value to be zero, because we actually have a dovage cabinet, so that the value is actually zero. And third, that's the requirement. I asked the value of the probability distribution of the values of the variable V is satisfied and required. That is at least a part that is equal or V is even higher. So that's the idea. That's why, based on what I can assume this. So that's why anchor in fact is important, because if there are hawkish members occupying pivotal positions in a cabinet, meaning that the dovage cabinet actually didn't have a landslide victory in party elections, otherwise they don't need any hawkish cabinet members occupying substantial positions. So this gives us sufficient evidence to support that the election result was closed, at least to certain extent it's closed. But it's not up to me to decide the structure as I said. So I just assume it's very closed. So now the question is, the fact is the value of V is one, a zero, sorry. So how can we intervene that? How to do that? So this is the values of actually that's hawkish cabinet or dovage cabinet where artists determined by the function of the values of HCM and V, that is the historical fact and the exogenous variable, and the function is like this. So only when the value of HCM and V are both one, FC, the value of FC is one, otherwise zero. But why I can get this result? Why this function is correct? So let's consider. First of all, FCM is one, is important for why this explains why it anchors the bad tracking process. You don't need to bad track further. So that's why it's there. It indicates that. So this is actually a combination of historical fact, that is there were hawkish cabinet members and the exogenous variable's value is zero as the default position. So this one is fine, right? So there is no hawkish cabinet, it's dovage. Oh, I forgot to say. For illusive purpose, all variables are minor. For illusive purpose, they are all minor. So this is first possibility and if there were no hawkish cabinet members, then N, the value of V is zero. So there is no hawkish cabinet. There were no hawkish, there was no hawkish cabinet. It's quite clear. But how about this one? Hawkish faction actually won. But there are no, there were no hawkish cabinet members occupied any substantial position. I think we don't need to consider this one. This is, this is the real oxymoron. So we don't need to consider this one. So then I explain why the function like this. So why expand this function? Because I think this is the formal structure between these variables, where you bad track to an angry bad. So we can have the first three conditions of the, the three conditions of the first set of all conditions. And the first one is that the, the three variables, the variables are sort of, it's very rewarding. So basically it's the bad track team counter fact that is HC. And the bad track historical fact is HCM, and the exogenous variable V. And they have the formal structure categorized by equation, by equation. So that's the first condition. The second is that the bad track team counter fact. The bad track historical fact, those two variables are endogenous. And only V is exogenous. So we can only thoughtfully intervene in V. And the third is that the, the requirement, the probability distribution requirement, because I want to talk in the ideal situation. So this has to be met. So this is the, the three formal conditions of the first set. So about bad track into a historical fact, bad track into an angry fact. So that's the idea. So the, the second part is about bad track into a counter fact. So I use this as an example that is bad track plus a fissuring element that is RS1 to a Hawkerich cabinet that is HC1. So using this part of bad track to explain the formal structure. So apparently a Hawkerich cabinet is not the only causal factor. For example, the economy after World War I has, had to be booming. Otherwise there, there, there, there was simply no money, no budget to, to faster the reality. So that's simply not possible. So there are some other causal factors related to the realization of a faster rearmament. So I can use you as standing for an auxiliary variable. It stands for the, all causal, the causal complex that representing all the causally relevant residue factors to the realization of a sufficient rearmament. So I just want a variable as such, represent the, the natural output or collective output. They have met, there must be a very complicated structure inside, but I don't need to discuss this. I just say they, okay, all of this you can use a variable representing. So now we can talk about the formal profile of this bad tracking. So the value of R as are determined by the functions of the values of HC and U. HC is Hawkerich, a selfish rearmament or insufficient rearmament, Hawkerich or Dovish cabinets and U is the auxiliary variable and the function is like this. Now again, why it is like this? So we can think about that, go through it. The first possibility is that it's a Dovish cabinet and the auxiliary variable's value is 1. It's a 10. So this one, if it's a Dovish, so it's 0, because it's fact. This part represents fact. Actually the cabinet was Dovish and actually there were no faster rearmaments. So this is fine. And the second one is that a Dovish cabinet and the auxiliary variable's value is 0, meaning that the needed causal complex wasn't there. There's no money, for example. So then no faster rearmament is reasonable. The third one is that there's a Hawkerich cabinet but still there's no money. So the meaning that the faster rearmament is still not possible, so the value of R is 0. And we can notice that when the value of the auxiliary variable is 0, there is no faster rearmament because there's simply no money. So apparently based on that, because I already assumed that this bad tracking is successful, it's sensible. So since it's sensible, so the default position must be U is 1. That must be the default position. Otherwise bad tracking, the whole bad tracking makes no sense. Why do you want to bad track from a sufficient rearmament to a Hawkerich cabinet when there's simply, there was simply no money to faster the rearmament process? If that's the case, why the bad tracking is sensible or successful? So because I assumed that the bad tracking is successful, so therefore the value of U must be 1. That is a fact. Has to be that. And given that the value of U is 1, the values of RS are actually determined by the values of HC, given by this setup. So that's why I say I don't only need one case, I just assume that it is a successful bad tracking and based on that I can formalize the correctness condition. So this is the extending of CC. Expanding of CC, we can add. So this is the subscript, stands for the siliquant, the silvery variable, which level. Since the bad tracking structure are similar, so actually we can put the allergy. The structure is exactly the same. So now we can have a second set of formal conditions. The first one saying is actually this, using cases is much more simple. The variable RS, that's the bad tracking counter fact. And the variable HC, that's bad tracked counter fact. And the auxiliary variable are all endogenous, or endogenous because they're coming from the fact part, not the kind of fact part. And the second one is that the auxiliary variable is fixed to be 1, the value of it is fixed to be 1. Meaning that it is, it represents a factual background condition. That's the factual background condition obtained. So it's bad. It's all there. It depends the character of the, the, the, the party faction. So that's it. So of course historical evidence might show otherwise. Say, ah, not the money was in there. My formulation is wrong. No, my formulation is still correct. What, what goes right is the assumption that assuming that the bad tracking is successful, they only show that the bad tracking is not successful. I need to find a new example. So although historical evidence might show otherwise, that doesn't mean my formulation has been failed. It's different. And finally, since the auxiliary variable value is set to be 1 or fixed to be 1, so the bad tracking, the value of the bad tracking variable is determined by the value of bad tracked variable. That is, rs values are determined by the values of hc. That is, severe armament and hawkish or doggish category. So that's the, the second set of three conditions. So new equation is out. New equation is out. Now we have this, we must have a seven condition. That is, using, using pictures. That is, there is no causal connection between auxiliary variables. That's one thing. Another thing is that there's no causal connection between exogenous variable and the auxiliary variable and the immediate incentive level. There's no causal connection here. So let's explain a little bit. Let's assume that urs is zero when v is one. Let's assume this. Assuming that there's some sort of causal connection. But the problem is this, that if urs value is zero, meaning that it fails to satisfy 2.2, that is, it's not a factual background condition. And I'll, I'll miss that. Assuming that the bad tracking is successful, the u, auxiliary variable has to be fixed to be one, representing the factual background condition. So this fail, formal condition 2.2. And can we say that it's possible that urs is one, when v is one. But the problem is, v is actually zero, meaning that urs is actually zero. And again, it fails to satisfy 2.2 condition. So there's no, no causal connection between the exogenous variable and the auxiliary variable and the immediate incentive level. It's not possible. So the second one is that the, about the auxiliary variable, whether or not they can have causal connection. So as said, the value of ua to b1 represents the factual background condition allowed to a bad track from a is zero to rs is one. To make sense the bad tracking. Because the money was there, so we can have this bad tracking. And apparently, urs, the value of urs is one did the same thing. Meaning that you can bad track from rs is one to rfc is one, that is hawkish cabinet. So assuming there's a causal connection between these two auxiliary variable, then it creates a unique problem. The problem is that there it did, now we have a full core possession in a bad tracking analysis. That is like this. Because the reason why it's of the bad tracking from a to rs actually rely on the reason why it's of rs bad tracking to fc. So this, there's a full character in a bad tracking analysis. And this is just wrong. So that's why there cannot be any causal connection between these auxiliary variables. So, as I said, history is a fact-based science. And you can see that these are all facts. The values are all fixed. They are all facts. So causal modeling is a sufficient tool to carve a sensible causal rule out of the channel. That is, history can use these facts to fix the background conditions. And try to use causal generalization or theories of rather than these beliefs and our historical facts as evidence to do the bad tracking and the kind of factor of bad tracking analysis. And causal modeling just helped to present the formal conditions. The correctness condition of a successful bad tracking, as I said, once this is set, we can explain each, for example, fc is zero as a historical counterfact. The historical consistency description is adequate, because the whole bad tracking is correct. So each value represents a historical counterfact or historical consistency. And any causal relationship, not here. Every causal connection here or causal description here, when one is at the immediate descent level, the causal description is also adequate. So as far as I can use causal modeling to construct the correctness condition of successful bad tracking, I can use them to explain the advocacy of historical consistent description, which are crucial to historian study. And based on that, I can further explain the advocacy of historical causal description. Thank you. Questions? I'll start. Could you bring back the, okay. So I see how you have to exclude a lot of relations for the thing to work, but do you not have to exclude, you know, a hawkish cabinet is inducing and on appeasement and dependency of the rearmament. So do you have to exclude that Harold? Oh, I see. You mean that whether or not there's a rearmament? In perfect. Because the cabinet is responsible to the anti-applicement. So directly or? Well, first of all, there's no, what's that? That's a very, that would be a very long cabinet for a very long time. No, that's hard to say. The causal relations cannot transfer. So the hawkish cabinet can only be the causal or the visual rearmament. So that's the idea, because it is using actually kind of factual and kind of factual cannot transfer the causal connection. I'll miss you, you want to use a kind of a kind of to tinkering with all the problems. Causality could be transitive, but counterfactual. Yeah, that would be a problem. So that's the feature, of course. Okay, okay. It's because it's a counterfactual reaction. Okay. Thank you. Other questions? Jonathan, why are you doing that? We can't call you a date and raise your hands. Let's call up that. Yeah, this is really just a curiosity question. So I wonder, I know there's some work in just in history, on counterfactual history. And so I wonder if you've seen some analogies between the formal constraints that you're proposing in this kind of approach and just like, when historians talk about doing counterfactual history, what kinds of guidelines do they give themselves for like, what's good practice? Well, I cannot explain this because I didn't actually spend time to study that, but I have been informed that somebody said, oh, I just read a book. Historians are actually doing that. And I have been told that the historian colleagues of philosophy say that they are doing things like this. So that's that much. And the funny part is this. I asked my historian colleague back to my university, they said, we never use counterfactual reasoning. We don't use counterfactual. We are doing historical study. We don't use counterfactual. So this would be kind of cool if one thing that this work could give, it's like, if they have these kinds of vague ideas about how to kind of structure this to get good inferences, and this could help them clarify, that would be cool. That's the idea. So I'm still working on this paper. Cool. And there are two other options. One is at Alexander Mark, trying to use standard Louisiana semantics to say that, okay, we can use that to talk about historical counterfactual reasoning. So whether or not they can build up the historical causal description or causality. But I have seriously done that. I think that's the main reason for my historian colleague saying that we never use counterfactual reasoning. Possible words. No, no, no, this word, this word, not possible. No possible word. Another is actually Glennon. Glennon tried to use ephemeral, ephemeral mechanism to explain historical causal descriptions. And I didn't spend much time just read one or two papers of his. But I still think that I think his account of mechanism has, why essential problem? Because he thinks that his account is superior than Woodward's, because he can explain the hierarchy. So that the lower level mechanism is the higher level causal generalization, some of the ideas, something like that. So you can use mechanism to explain causal generalization. And when I taught this, I and my students, our question is just that what's going to happen at a button level? You must have a button level. That there's no nothing causal going on within the mechanism. That has to be the thing. So basically, I think that the Glennons that come, I don't think he can explain historical causal descriptions, although I don't have any. But I know that at least there are two different approaches, not just this one. They are two different approaches. So I will say I probably understood about 20% of what you're doing, but because of the lectures, because I couldn't keep up with all the articles, but for whatever this is about, when the British did re-ar, they deficit-financed it. So the economy did not have to be very good. Right. Yes, of course, of course. But has to be good, to be sufficiently fairly good. We cannot be bad. It has to be to some extent. But it's why you know it's terrible when it's, but it's why it's an exogenous variable. It's anything that will create the necessary condition for the global economy. Well, the exogenous- So the economy, Dunkirk happens, Britain is being bombed, they're losing their colonies left and right from, because of, I mean, this is, they are desperately debt-financing. Anything that they're doing, just if I disagree with you. That's probably true, because I'm not an analogy story. But the thing is, I just assume that this bad tracking is successful. And I say, if given that it's successful, the formal structure has to be like this. Because if the value is zero, the bad tracking just doesn't make any sense. So you are actually saying that the bad tracking is not that successful, which can be true, and likely to be true, but I just assume it to be true after or it's just an example. It's just a case. I just, I want the formal structure. This means that I would defend it. Daniel is arguing for a more complex graph, but you're not saying that the condition of the graph is false. Actually, actually, the paper I discussed a bit more. For example, whether or not we can have a bad tracking about this one, that's one possibility. Can we have multiple, multiple preconditions, not just one precondition, whether over determination and preemption might cause any problem at this graph. So that's why I remember when I said this is the basic ideas. Because I think that I present this once. I spent 30 minutes to explain this. As you said, oh, I just don't remember that. So I didn't. But, but, but for the record, let me prove this to you. It's okay. It's okay. I'm not joking. It's okay. I just don't have time. Oh, sorry. And really fast. So this model can be used to understand, for example, the causal history of a functional trait. Because there is this discussion about functions, what is a functional effect? Are we, they say that a functional trait is actually the selected effect in the history of the system. So they talk about some kind of backward causation to talk about functionalities in organisms. What's the problem? They don't have a model to to put this history. My question is, what are the limitations of variables to know? How many variables we have to know to make a history? Because in history, in general, we have knowledge of a bunch of variations. But in the biological evolution of a system, we don't have, for example, much knowledge because there is a lot of things that happen, mutations and external boundaries, chains that we don't know what's happened. We know, for example, that there is a kind of animal that had some traits and then another animal that had other traits. But we would say, for example, that a trait that we have now is a functional trait because it's evolution of history. And that is a backward causation history. Well, first of all, I'm not talking about backward causation. And second of all, as I said, I can only talk about human history. Human history. Just human history. Because the thing, the whole thing is that I think the last time I did that. So as I said, there's a chunk of historical facts, a lot of historical facts. Just based on those facts by themselves, it's very difficult to develop a causal, to explain the causal loop. So this model, the bad tracking analysis, is bad tracking, not backward causation. Bad tracking analysis can help to establish the analysis. And then causal modeling can help to establish the formal conditions to explain why. At each level, we need historical facts as evidence. And this is a chunk, this is a causal chunk that might be very, very complicating inside. And history, luckily not me, history has to look into to establish, because I think that even if they wanted to establish a bad tracking analysis here, for example, I think the formal conditions are exactly the same. So that's why I just assume that this is a simple but successful bad tracking, and try to construct the formal conditions of the correctness of this bad tracking. And hopefully, this can keep you see. So that's why I think about discuss the orientation preemption, because they might feel that the formal conditions are all satisfied by the causal description is not, that's already the main issue of preemption. And try to explain why this won't cause any problem. So that's the thing. So it's all about human history. And there's something I forgot to say. Oh, okay, there's one thing I have to say. Yeah, one thing that, of course, there must be a limit. You cannot bad track forever, right? I read, I think it's, for me, they said to explain the booming of Western civilization in the 19th century. That's because the Greek won the war of the state war. I think, what? There's other things. It couldn't possibly construct any reasonable bad tracking. It's 2000, more than 2000 years. That's not possible. Of course, it's with the scale limit. But there's a, there's a huge solution. The actual limit of the scale has to be decided by historians. I can't decide that. I can't say that. Oh, you can't bad track to more than five months. How can I say it? For 10 years, I can't say that. Historians.