I knew it was going to be a bad day when I opened my email this morning to a message from a colleague that linked to a new study in Nature: “Civil conflicts are associated with the global climate.” (the actual article is paywalled). Well, that is assertive . . . especially because despite similar claims in the past, I have yet to see any study make such a definitive, general connection successfully. Look, the problem here is simple: the connection between conflict and the environment is shaky, at best. For all of the attention that Thomas Homer-Dixon gets for his work, the simple fact is that for interstate conflict, there are more negative cases than positive case . . . that is, where a particular environmental stressor exists, conflict DOES NOT happen far more often than it does. Intrastate conflict is much, much more complex, though there are some indications that the environment does play a triggering/exacerbating role in conflict at this scale.
Sadly, this article does not live up to its claims. It is horrifically flawed, to the point that I cannot see how its conclusions actually tell us anything about the relationship between El Nino and conflict, let alone climate and conflict. Even a cursory reading reveals myriad problems with the framing of the research design, the regression design, and the interpretation of the regression outputs (though, to be honest, the interpretation really didn’t matter, as whatever was coming out of the regressions was beyond salvation anyway) that lead me to question how it even got through peer review. My quick take:
Let’s start with the experimental design:
… We define annual conflict risk (ACR) in a collection of countries to be the probability that a randomly selected country in the set experiences conflict onset in a given year. Importantly, this ACR measure removes trends due to the growing number of countries.
In an impossible but ideal experiment, we would observe two identical Earths, change the global climate of one and observe whether ACR in the two Earths diverged. In practice, we can approximate this experiment if the one Earth that we do observe randomly shifts back and forth between two different climate states. Such a quasi-experiment is ongoing and is characterized by rapid shifts in the global climate between La Niña and El Niño.
This design makes sense only if you assume that the random back-and-forth shifting did not trigger adaptive livelihoods decisions that, over time, would have served to mitigate the impact of these state shifts (I am being generous here and assuming the authors do not think that changes in rainfall directly cause people to start attacking one another, though they never really make clear the mechanisms linking climate states and human behavior). The only way to assume non-adaptive livelihoods is to know next to nothing about how people make livelihoods decisions. Assuming that these livelihoods are somehow optimized for one state or the other such that a state change would create surprising new conditions that introduced new stresses is more or less to assume that the populations affected by these changes were somehow perpetually surprised by the state change (even though it happened fairly frequently). After 14 years of studying rural livelihoods in sub-Saharan Africa, I find that absolutely impossible to believe. Flipping back and forth between states does not give you two Earths, it gives you one Earth that presented certain known challenges to people’s livelihoods.
To identify a relation between the global climate and ACR, we compare societies with themselves when they are exposed to different states of the global climate. Heuristically, a society observed during a La Niña is the ‘control’ for that same society observed during an El Niño ‘treatment’.
No, it is not. This is a false parsing of the world, and as a result they are regressing junk.
This is not the only problem with the research design. Another huge problem with this study is its treatment of the impact of ENSO-related state changes on people. These state changes in the climate do not have the same impact everywhere, even in strongly teleconnected places. The ecology and broader environment of the tropics is hardly monolithic (though it is mostly treated this way), and a strong teleconnection can mean either drought or flooding . . . in other words, the el Nino teleconnection creates a variety of climatological phenomena that play out in a wide range of environments that are exploited by an even larger number of livelihoods strategies, creating myriad environmental and human impacts. These impacts cannot be aggregated into a broad driver of conflict – basically, their entire regression (which, mind you, is framed around a junk “counterfactual”) is populated with massively over-aggregated data such that any causal signal is completely lost in the noise.
Most reasonable approaches to the environment-conflict connection now treat environmental stresses as an exacerbating factor, or even a trigger, for other underlying factors. Such an approach seems loosely borne out in the Nature article. The authors note that in the “teleconnected group, low-income countries are the most responsive to ENSO, whereas similarly low income countries in the weakly affected group do not respond significantly to ENSO.” This certainly sounds like a broad stressor (state change in the climate) is influencing other, more directly pertinent drivers of conflict. But then we get to their statement of limitations:
Although we observe that the ACR of low-income countries is most strongly associated with ENSO, we cannot determine if (1) they respond strongly because they are low-income, (2) they are low income because they are sensitive to ENSO, or (3) they are sensitive to ENSO and low income for some third unobservable reason. Hypothesis (1) is supported by evidence that poor countries lack the resources to mitigate the effects of environmental changes. However, hypothesis (2) is plausible because ENSO existed before the invention of agriculture and conflict induces economic underperformance.
Even here, they have really oversimplified things: the way this is framed, either the environment causes the conflict (pretty much established by the literature that this is not the case), the environment causes economic problems that cause the conflict, or it is something else entirely. Every other possible factor in the world is in that third category, and most current work on this subject concentrate on other drivers of conflict (only some of which are economic) and how they intersect with environmental stresses.
This paper is a mess. But it got into print and made waves in a lot of popular outlets (for example, here and here). Why? Because it is reviving the long-dead corpse of environmental determinism…people really want the environment to in some way determine human behavior (we like simple explanations for complex events), even if that determination takes place via influences nuanced by local environmental variation, etc. Environmental determinism fell apart in the face of empirical evidence in the 1930s. But it makes for a good, simple narrative of explanation where we can just blame conflict on climate cycles that are beyond our control, and look past the things like colonialism that created the foundation for modern political economies of conflict. This absolves the Global North of responsibility for these conflicts, and obscures the many ways in which these conflicts could be addressed productively.
re: the title of your post, I think it got through peer review because it was published in Nature, and not the APSR
http://themonkeycage.org/blog/2011/08/03/a-modest-proposal/
It does seem that Nature is engaging in a bit of intellectual ambulance chasing, no?
It seems that the conflict/crisis narrative has been gaining more traction in recent climate change news. Which is not necessarily a good thing… I’m really interested in how the crisis framing of climate change affects the international agricultural aid/development agenda. Shades of the Green Revolution and population bomb narrative?
Marci:
Good question – I certainly see the climate and security stuff rising in everyone’s consciousness, and I think it can be mobilized in any number of ways. Right now, I think most folks I encounter in all sorts of institutions seem to focus on conflict as a development challenge – you can’t have development in conflict-ridden areas – and that is how it enters the development agenda. However, it could leak into other conversations in any number of ways . . .
You say “These impacts cannot be aggregated into a broad driver of conflict – basically, their entire regression… is populated with massively over-aggregated data such that any causal signal is completely lost in the noise.”
But that’s wrong. They do successfully extract a signal. Both in the main result, in the monthly data and in their hierarchical model (the one you cite regarding low income countries). They are simply estimating an “average treatment effect.” Of course not all countries respond identically, but this is a standard and useful assumption when we only have datasets of limited size. Its fair to point out that all countries do not respond identically to El Nino, but the authors do verify that the average effect in the tropics is warming and drying (in their online appendix).
Everything the authors have done here is standard econometrics. I’m not sure why you are so dismissive of it. You may not like the use of statistics generally in development work if you think statistics are incapable of capturing important phenomena. That’s okay and you’re entitled to that opinion. But if that’s your concern, you should just state it, since your critique of how the statistics are carried out in this paper is simply wrong. The authors were, by all standards, insanely cautious and rigorous with their analysis. The online appendix is 50 pages long, almost entirely dedicated to checking every single one of their statistical assumptions and techniques. That kind of quality analysis is probably exactly why it made it through peer review.
Thanks for your comment. Solomon just left a long response in the comments, and I responded to him at length. I hope my response there clarifies some of the issues you raise here. But also let me be clear, I am not dismissive of econometric analysis in and of itself. I believe in using the right tool for the job, whatever that tool is. I am not actually contradicting how they ran the regression . . . my critique runs at a much more meta level: they did not have a deep enough understanding of the issue they were trying to interrogate to set up a meaningful analysis. I stand by that. I have no doubt that, within the bounds of this study, they were cautious and did a good job of checking their statistical assumptions – but that is not the critique I am leveling. I am arguing that they failed to check their broad assumptions about how the world works that might inform the very questions they were asking, and how they went about asking them.
I was with you until the last two sentences. My sense is that much of the work on climate and conflict is driven by people who want to highlight, not bury, how poorer countries are being screwed by the lifestyles of the rich and powerful. It’s precisely because climate change is in our control that this research agenda has attracted so much interest.
Perhaps it was my phrasing. I agree that a lot of work on climate and conflict does try to make this connection (though not all of it does) . . . but my point in those last two sentences was simply to say that there is an implicit message to the study at hand (at least for those who look for simple causes and solutions): these conflicts are driven by an environmental factor (in this case climate oscillations) and therefore we don’t have to look at the things that are triggered by these environmental issues. You have a good point, though – we could still see this as a call to address the root causes of climate change (but of course this is ENSO, not climate change). However, the article certainly seems to rule out addressing the local issues on which these environmental stressors play out.
I am the lead author on the paper criticized in this post. A colleague asked me to explain my response to this critique, so I have cut and pasted my response to him below. I have tried to address each substantive comment directly.
I will also point interested readers to a website where we have made the detailed Supplementary Information Appendix (detailing and stress-testing our statistical analysis as well as their underlying assumptions) and replication data/code available for free:
http://blogs.cuit.columbia.edu/smh2137/research/publications/
On that site, there is also a video and Google Earth file that will let readers unfamiliar with ENSO get an intuitive feel for the structure of the global teleconnections analyzed in the paper.
I welcome clarifying questions about this response, however I will not reply to any comments that contain only disparaging remarks and lack substance, although I trust that Edward would have moderated out those comments anyway.
-Solomon Hsiang, Princeton University
————–(1)
“For all of the attention that Thomas Homer-Dixon gets for his work, the simple fact is that for interstate conflict, there are more negative cases than positive case . . . that is, where a particular environmental stressor exists, conflict DOES NOT happen far more often than it does.”
This is irrelevant to our study. We were clear to only examine intrastate conflict because the overwhelming number of modern conflicts around the world are intrastate, rather than interstate.
This graph generated by the group at the Peach Research Institute Oslo demonstrates this point very clearly:
http://www.prio.no/sptrans/-193189583/Graph%20-%20Conflicts%20by%20Type.pdf
————–(2)
“This design makes sense only if you assume that the random back-and-forth shifting did not trigger adaptive livelihoods decisions that… served to mitigate the impact of these state shifts…”
This criticism does not invalidate our research design in any way. It is clear that individuals on the ground adapt their lifestyles to environmental changes. This is, in fact, exactly what the paper is about. Individuals choose actions from a portfolio of possible choices, one of which is to engage in conflict. The study is trying to understand whether choosing to engage in conflict is a “livelihood decision” that individuals in modern societies select more often when El Nino events occur. Our findings tells us that for some reason, people’s willingness to engage in organized violence changes when the global climate changes. One hypothesis is that perhaps “predation” (i.e. the forceful extraction of property from others) is a form of “adaptation” to climate changes. A nice paper explaining this idea (not focused on climate per se) is Dal Bo and Dal Bo (2011, Journal of the European Economics Association):
http://onlinelibrary.wiley.com/doi/10.1111/j.1542-4774.2011.01025.x/full
(working paper for free here:)
http://www.econ.brown.edu/fac/Pedro_Dal_Bo/workerswarriorscriminals.pdf
with nice empirical tests like this one by Dube and Vargas (Revise and resubmit at Journal of Economics Studies):
https://files.nyu.edu/od9/public/papers/Dube_Vargas_commodities_conflict.pdf
————–(3)
“Assuming that these livelihoods are somehow optimized for one state or the other such that a state change would create surprising new conditions that introduced new stresses is more or less to assume that the populations affected by these changes were somehow perpetually surprised by the state change (even though it happened fairly frequently).”
This is incorrectly conflating two issues. (1) whether individuals are surprised about the *timing* of an El Nino event (2) whether individuals are surprised about what an El Nino climate looks like for them [and how they can optimize their life under those conditions].
We actually would expect that individuals on the ground are familiar with what El Nino conditions look like and what kinds of choices will optimize their livelihoods under some given ENSO conditions (because each state occurs frequently). However, we do not expect that most individuals know a year in advance whether the next year with be an El Nino year or not. For our research design to be valid, all we need is for individuals on the ground to have difficulty anticipating *when* El Nino events occur. The fact that they have observed an El Nino before does not threaten the scientific validity of the research design.
————–(4)
“Flipping back and forth between states does not give you two Earths, it gives you one Earth that presented certain known challenges to people’s livelihoods.”
This is clearly correct, but it does not invalidate the research design. To see why, imagine that I want to study whether New York City subway closures make people late to work. I could look at historical data on whether a random subway closure occurred and then look at whether people were late to work on that day. Obviously people know that subway closures will occur sometimes, but if the timing of the closure is hard to predict, many people will plan their commute assuming that the subway will be running most of the time. Then, on the days when the subway is closed, they will show up to the station, find that it is closed and manage to find some other transportation to work. They may be annoyed and angry, and importantly for the study, late to work. But if you had asked them earlier that day (before they left their house) “is there a chance the subway will be closed today?” any experienced New Yorker would say “of course.” The fact that subway closures are expected by all commuters, in a statistical sense, doesn’t prevent those closures from impacting people’s lives or from allowing us to research whether a closure will make people later to work on average.
The notion of “two earths” is simply a heuristic used to help people understand how time-series data on the global climate can be used to learn something about how societies relate to it. In the subway study above, an analogy would be to “picture two versions of New York City, one with and one without subways”, and then ask “how long does it take people to commute to work in each situation?” Of course studying subway closures would only be an approximation of this “perfect experiment,” but it may be the best we can do given the constraints of the real world (i.e. we can’t build a “control” NYC with no subways).
Probably the most famous paper describing how quasi-experiments can be used, and under what assumptions they are valid substitutes for actual experiments, is Holland (1986, Journal of the American Statistical Association):
http://www.jstor.org/stable/2289064
————–(5)
“[T]he el Nino teleconnection creates a variety of climatological phenomena that play out in a wide range of environments that are exploited by an even larger number of livelihoods strategies, creating myriad environmental and human impacts. These impacts cannot be aggregated into a broad driver of conflict…”
Again, it is true that El Nino teleconnections create a variety of climatological phenomena, but this fact does not invalidate our study in any way. Our study examines whether El Nino, as a state of the global climate that has many dimensions, changes the risk of conflict on average. This criticism is analogous to saying (with regard to the hypothetical subway study) that since people travel different distances to work, work at different times of day, and some own a bike/car, it is impossible to estimate whether subway closures make people later to work on average. Of course a subway closure will affect different people differently, and future studies might try to look at the detailed effects on people who live in Brooklyn compared to Manhattan, but it doesn’t mean that the *average* effect is incorrectly measured.
As a side note, in the paper we do document that El Nino leads to hotter and drier conditions on average in teleconnected countries (see Supplementary Table 2). This is well established in the climate sciences, with a mechanism that was explained clearly by Chiang and Sobel (2002, Journal of Climate):
http://www.ldeo.columbia.edu/~sobel/Papers/chiang_sobel.pdf
A short summary of their paper: The release of thermal energy from the tropical pacific generates a Kelvin wave that carries this signal around the tropics. This increases the stability of the atmosphere throughout the tropics, slowing both dry and moist convection. This generally leads to a reduction of rainfall and to an increase of surface temperatures (because surface air must become relatively warmer before it can rise in the relatively more stable atmosphere).
It is true that El Nino can induce flooding in some regions, but that is generally limited to coastal regions where increased ocean evaporation (due to warming) leads to increased local rainfall. In addition, in our Supplementary Information we look to see if any countries become systematically cooler during El Nino (i.e. we look for “negatively teleconnected countries” using the same method we use in the main text) and find that only New Zealand, Fiji and the Solomon Islands would satisfy this criteria (see Supplementary Figure 5).
————–(6)
“[B]asically, their entire regression… is populated with massively over-aggregated data such that any causal signal is completely lost in the noise.”
This claim is simply false. It is true that our data is the most aggregated social data that we have ever seen analyzed (it summarizes the conditions for half the world’s population) and often aggregation makes signal detection difficult. However, this only makes the fact that we can extract a signal from the noise that much more remarkable. Despite everything else in the world that’s going on, we can observe a signal from ENSO loud and clear. And not only do we observe a single correlation, but we observe four patterns all of which point towards the idea that ENSO affects conflict:
1 – ENSO and conflict are correlated annually in the tropics where ENSO’s influence is large.
2 – ENSO and conflict are no correlated in the weakly-affected countries where ENSO’s influence is small.
3 – Within the tropics, ENSO and conflicts are correlated on a monthly basis, with additional conflicts beginning only during the period when El Nino is the dominant phase.
4 – We find that poor countries are statistically more vulnerable to ENSO in terms of conflict. (If we found the reverse, we would be very hesitant about interpreting our results causally.)
If Carr’s above quote meant to say the correlation we observe is just spurious (i.e. generated by random noise) he would have to explain how random noise would simultaneously generate all four of these structures in the data with any reasonable probability.
————–(7)
“This paper is a mess. But it got into print and made waves in a lot of popular outlets (for example, here and here). Why? Because it is reviving the long-dead corpse of environmental determinism… people really want the environment to in some way determine human behavior (we like simple explanations for complex events), even if that determination takes place via influences nuanced by local environmental variation, etc. Environmental determinism fell apart in the face of empirical evidence in the 1930s. But it makes for a good, simple narrative of explanation where we can just blame conflict on climate cycles that are beyond our control, and look past the things like colonialism that created the foundation for modern political economies of conflict. This absolves the Global North of responsibility for these conflicts, and obscures the many ways in which these conflicts could be addressed productively.”
This is an incorrect interpretation of the papers findings. We worked very hard to be clear with the press that our findings do not support “environmental determinism,” and this showed through in the reporting. For example the Nature News article states “El Nino events, [Solomon] adds, are by no means the sole factor leading to conflict.”
http://www.nature.com/news/2011/110824/full/news.2011.501.html
In press briefings, we have explicitly stated that while ENSO influences 20% of conflicts, this means that 80% of conflicts seem unrelated to ENSO. This clearly suggests that many things other than climate, such as politics, economics and social conditions, influence conflict. What our study shows is that along with these other factors, climate also plays an important role.
Solomon:
First, let me thank you for responding, and for posting a response here. Please be assured I will not tolerate responses without substance – I want this to be a site where ideas get discussed, and shouting is minimized. I moderate the comments here. This goes doubly, given your choice to post a response here – I would like to honor that effort with a real conversation.
To work through your response:
1) Yes, you all did stay intrastate, so Homer-Dixon is irrelevant in some ways. On the other hand, any discussion of conflict and climate change is engaging a discourse that has been heavily shaped by that work, and so while your article was not guilty of the same issues, I think that a lot of readers of the piece (and most certainly those who read the popular interpretations of the piece) were thinking along the lines of “water wars,” etc. To be clear, I am not suggesting that you and your colleagues were replicating Homer-Dixon, or choosing to mobilize that discourse for some sort of an advantage, but that we really do have to address that larger discourse as we move forward on the climate and conflict (and really the environment and conflict) front.
To address your concerns in 2 and 3, I need to go back to my main concern: I don’t think that you all understand livelihoods very well. I am not saying that to be disparaging – there are a hell of a lot of things in this world that I do not understand very well, and I am certain there are things you and your co-authors understand much better than I do. But livelihoods is something that I feel quite certain I understand better than you all, an argument I am happy to support via my CV or an offline discussion.
For 2), a constructive suggestion: never refer to conflict as a “livelihoods decision,” and for the love of God please don’t call predation an “adaptation” I know you used quotes, but the anthropologists and a lot of geographers will dismember you for it. I take the point you were trying to make, but please say it differently in the future lest an ungenerous audience start screaming . . . though I am sure you don’t mean to, you risk sounding like someone who attributes the behavior of the global poor to animalistic instincts in a manner that we don’t see here in the Global North. To address the rest of my concern with 2), I need to address 3) and 4).
For 3), you argue that timing is really the key issue. I agree that timing is more important than the state itself, as people will already recognize different states in their livelihoods. However, you presume that timing matters in the same way in all cases, and it does not. Some crops are much more vulnerable to shifts in precipitation than others, some watersheds handle excess rainfall better than others, etc. So the timing of a state change matters more or less depending on where you are, what the local environment looks like, and what local livelihoods look like. You have not controlled for any of this in your analysis. As a result, you have aggregated populations that are highly sensitive, somewhat sensitive, and largely insensitive to state changes and their timing. This aggregation makes it impossible to really get at the causal linkages between environmental change and conflict. I would be very interested to see what happens if you decompose your data, even coarsely, for local ecology and livelihoods. Those results, I think, could start to get quite interesting.
I think my big issue with your analogy on 4) is that, at least in much of rural sub-Saharan Africa, livelihoods are far more hedged than your example would suggest. Missing a subway has a cost, but not a huge one. A failed crop kills people. Rural livelihoods are deeply hedged for this risk, and therefore as they plant they do not assume things will go according to plan. There is a huge lit on this, including a lot of stuff that misinterprets this hedging as sub-optimal decision-making. This deep hedging erodes your fundamental assumption about people’s behavior here, and I think presents a real challenge to the “two Earths” heuristic.
5) See my point on 3) – while this may be fine statistically on a generic dataset, when talking about livelihoods and the environment (which, fundamentally, you are) you simply cannot aggregate this data and talk about what happens on average, because this is not getting you meaningful information. You need a theory of the connection between environmental change and conflict that can underpin this study . . . or you need to disaggregate the data as I have suggested above and see what sorts of patterns emerge, and theorize from that as you re-aggregate (though I suspect you will find that you cannot, in fact, meaningfully re-aggregate). Also, please note that variations in ecology are such that general framings of climate under ENSO are still playing out in deeply complex, variable ways in particular places . . . and you have not controlled for this in your analysis.
6) My responses 1) In my opinion, you have not adequately addressed what an impact of ENSO looks like specifically, and therefore cannot say if the influence is in fact large. All you can say is that there is a strong teleconnection, but the climate outcomes of that teleconnection is filtered through local environment and ecology into local livelihoods, all of which are highly variable in terms of their sensitivity to these impacts (see points 3) and 4) above). As a result, I have little confidence in this correlation. 2) Same point as 1). 3) This is hugely important – the idea that ENSO and conflict are correlated on a monthly basis doesn’t really make sense, unless you are willing to suggest that the climate has a direct, unmitigated impact on people’s behavior (i.e. environmental determinism). An ENSO event takes time to filter through the environment, ecology and livelihoods to create impacts on people that might then be acted upon. This can take a month or more. So a month-to-month correlation actually is a problem, not a strength. 4) We already know that poor countries are more vulnerable to conflict in general – this is well-established, though again the causal links are pretty complex. So you are going to find higher rates of conflict in poorer countries, with or without ENSO – it is entirely unsurprising you would have a higher correlation here, even considering all of the issues I outlined above. It might be the one variable that actually holds here.
7) Let me be clear (and by way of doing so offer an apology): I do not think that the authors of the paper consciously set out to revive environmental determinism – if that is how people read this, I apologize, as that was not my intent. That said, however, there are many places in the paper where there are some uninterrogated assumptions that go by (such as my response to point 3 under 6) above) which enabled such interpretations. More to the point I was trying to make, the popular pickup of the article was all about a neo-environmental determinism. We already know that climate plays a role . . . as a stressor on existing issues. I am not convinced that your study even really reinforces that message, mainly because of the issues I have raised in my post and in my comments here. I think that disaggregating the data, and getting more information on environment and livelihoods might allow for such a contribution. You all did put disclaimers in . . . but I disagree that these were reflected in the press. I cannot blame you for the press – that is completely unfair. All I can do is ask that you continue to explain what you think you found, as opposed to what people want you to have found (though I remain unconvinced that you have found it here).
Interesting back-and-forth between you and Hsaing. But can you expand on your point about “livelihoods”? Your argument seems to boil down to:
1. The authors don’t understand livelihoods in the way that you do.
2. Some anthropologists and geographers might not understand this work or might understand but try to put words in the authors’ mouths.
My read is that the second point is largely about semantics, and while it may be important for communication across disciplines it is not about substance and does not invalidate the research. But I’m not sure what you’re on about with regards to livelihoods. What is it specifically that the authors don’t understand, and how does it affect their analysis (even in a “meta” way)?
Joe:
Thanks for your comment. Starting with your second point, I am not so worried about the semantics issue that I raised in my response to Hsaing – I intended that to simply be cautionary, so they could avoid cross-disciplinary problems. But I think that there are a lot of semantic things happening in that argument that inadvertently point toward environmental determinism as an explanatory framework, and I am certain this is why the article got such pickup in the popular media. Those sorts of semantics are very, very important, but you are correct – semantics alone do not invalidate an argument.
Your pointing out that I have not been clear about how they misunderstand livelihoods is very helpful, as it helps me to see where I am talking past a lot of folks. I think I am guilty of being way too close to the subject, and therefore not expressing myself very well. However, I can’t do it in a reply. I will work up a short post that links to some of my academic work that lays out what I mean, and why it matters to their analysis, shortly (slightly swamped for the next day or so).