All of the uncertainty around COVID-19 means that we'll build whatever complex models we want, but then just stare blankly and wonder how the hell to choose values for various coefficients.
You can drive the suggested analysis here as far as you want, but ultimately, you will lift your head and realize it's all useless for actual policy making wrt COVID-19.
The ultimate answer is that there is no ultimate answer. Abstract mathematical modeling for these sorts of questions is therefore mostly useless. Useful answers come from going in and dealing with lots of messy particularities in a very messy reality.
I like to think that this problem with quantitative rationality, where you build a model about one thing and get a very precise answer, but then realize your answer is really just another question, is exactly what's being alluded to in "42".
(e: moved this sentence to the bottom of my comment because it seems to be causing some distracting hostility: "This sort of thought process is exactly how you go about convincing a wise person to just give up on the whole hard core rationality project. It's an unintentional strawman.")
Generally speaking, if you are going to offer criticism, try and offer a solution or ask a question that would address that criticism.
All I see here is "The value of human life is incalculable, so just give up."
Am I wrong? What am I missing?
We don't have infinite resources or infinite capacity to distribute them. That means we need to make hard decisions about things like medical care. That is reality. If I get some kind of crazy-aggressive brain cancer that costs a billion dollars to treat, I fully expect to die.
The best way to make those decisions is to have some kind of model that can (a) be readily understood; and (b) helps us make those decisions fairly and with as little unintentional bias as possible.
There will always be bias -- it's part of the human condition -- but if we can be transparent about our biases and willing to discuss them, we open the door to improving our decision-making systems later on.
With that in mind, what would you suggest as a model for determining, say, how we budget for healthcare?
> if you are going to offer criticism, try and offer a solution or ask a question that would address that criticism.
My solution is that the solution probably isn't useful so don't bother. Seriously. The game of assigning quantitative value to lives is almost never going to help you make better decisions.
> All I see here is "The value of human life is incalculable, so just give up."
Nope. It's exactly not that! I don't think it's incalculable. You can definitely choose a way to calculate it. Precisely, even. But it's going to be some complex function. And you're going to plug that function into another model. And when you do that, one of two things will happen:
1. you'll ground out in a deontology because you choose extreme values, or
2. the uncertainty in the other parts of the model will dominate.
In COVID-19, #2 is what happens.
> With that in mind, what would you suggest as a model for determining, say, how we budget for healthcare?
I suggest that any reasonable model would look more like an entire industry of mathematicians thinking about edge cases than what we would recognize as a single mathematical model. And even then the answer would probably suck. And definitely nowhere would you find it useful to have a quantitative assignment of lives vs. money, because anywhere you try to use that is going to have a ton of uncertainty that dominates the useful range of possible values. Which pretty much comports with my original comment that "there is no universal answer". I'm not an actuary or policy analyst, though.
Not the OP, just wanted to add that not all problems have solutions, especially the hard problems. Believing otherwise is very "technicist" (for lack of a better word), is believing that the "world" is a closed-system that "moves"/behaves in a predictable and computable manner. It doesn't.
It's easier for us as a species to say what a star billions of light-years away has been "doing" since its "inception" and the most probable way in which it will "die" than to say what the 5-year old kid next doors is going to do in a year or two, heck, what he's going to do tomorrow.
Sure, so give me a number. Eventually somebody[1] has to make a high-level decision on which strategy to pursue. And they have to give directives that can be followed by people who don't know your entire thought processes. That's what it means to be a bureaucracy. So which is it, will you choose an imperfect quantitative model where some numbers are pulled out of your ass? Or will you just let every low-level functionary pull entire decisions out of their individual asses and hope something resembling a coherent strategy will emerge?[2]
[1] Well, ideally. The US response to covid was pretty much [2]. If it wasn't clear, [2] was a rhetorical question and you should not do that.
42? I'm genuinely not sure what number you're asking for.
> Eventually somebody[1] has to make a high-level decision on which strategy to pursue.
Yes, of course, but that decision making doesn't require a quantitative model assigning value to life (or not). You can make a lot of very well-informed decisions without ever thinking about that particular aspect of optimization theory with any particularly high level of detail. "Aggressively avoid protracted exponential growth" doesn't require knowing anything about the value of life relative to other things.
Mathematical modeling can be useful for responding to covid-19. However, mathematical modeling for this question is not useful, because your answer gets plugged into yet another model where there's too much uncertainty for your answer to the have any quantitatively justifiable impact on decision making.
To give a toy example, spending a bunch of time figuring out the value of $c$ in $f(x) = x^n + c$ is a waste of time if you don't know much about the value of $n$.
All of the uncertainty around COVID-19 means that we'll build whatever complex models we want, but then just stare blankly and wonder how the hell to choose values for various coefficients.
References? COVID-19's behavior has been quite simple so far. Exponential growth with variations. It's more predictable than the weather next week, overall.
The American curve and European curve are somewhat different, yes but the differences can be explained by the total fucked-up the American virus control efforts have been.
Oh, and the American failures were accompanied by complaints about uncertainty and how much virus control efforts would cost. Now, we will find out how much a lack of control is going to cost.
"COVID-19's behavior has been quite simple so far."
What do you mean with behavior? People are still discussing how it spreads.
The latest news is it travels as aerosol and that in badly ventilated rooms keeping 1.5m distance is not going to help much. And also that the amount of virus particles you inhale determines how sick you get.
So it is only predictable when we know how it is growing exponential.
The still ongoing discussion about how it spreads creates uncertainty.
There is far less uncertainty around COVID-19 now than there was at the beginning of the year. It is possible to see which countries are successfully tracking and tracing to eliminate, which are entering second waves and need to mitigate, and which haven't even finished their first.
With respect to specific modelling efforts policy makers sometimes act as antagonists because, without joint modelling of all the concerns they are trying to balance, they are forced to rely on their intuition. This leaves society particularly exposed to the character traits of our leaders, because it is they who end up modelling the interaction effects (in their heads).
For example they might claim in the early stages of an epidemic that we actually just have linear case growth. It is then up to the modeller to try to justify why that isn't the case, which in limited-information regimes can be near impossible in any rigorous sense, for the reasons you describe.
However, assumptions are important, and showing decision makers what might happen given certain assumptions (and providing a justification for those assumptions), which is part of what a model does, can cause them to reassess their current thinking, and it is for this reason that these kinds of models can remain useful, even if it is not possible to reasonably distinguish between some set of models given available data.
Returning to the epidemic example, justifying why exponential rather than linear growth is likely in the early stages of spread is something that can be done on an intuitive level, and explaining this to policy makers, and showing them the range of possible outcomes that are consistent with current (at the time limited) knowledge, can be extremely powerful.
To add to the above, the distribution of population-level COVID outcomes has fat tails (the system evolution is non-ergodic), so computing point estimates with ensemble average “expected values” is junk modeling. And computing the (correct) time averages requires owning up to the uncertainty and fatness of tails (which is a very hard thing to model), so people have the lamppost bias of doing the easy (but useless) calculation.
An alternative way to explain the meaning of “large uncertainties” is that you can fudge input numbers slightly to get whatever output decision you want, so whoever is making the policy decision is likely to keep fudging numbers till the outcome “feels right”. At that point, the modeling is just dangerously misleading pseudo-science.
An intuitive approximation of the decision based on time averages is to “play it safe” in the presence of fat negative tails. “How safe tho?” is a very hard question to answer because of the uncertainties involved (as parent post explained).
> Studies that estimate the monetary price we are willing to pay to save a life have long shown puzzlingly great variation across individuals and contexts. Perhaps in part because the topic is politically charged. Those who seek to justify higher safety spending, stronger regulations, or larger court damages re medicine, food, environmental, or job accidents tend to want higher estimates, while those who seek to justify less and weaker of such things tend to want lower estimates
If lawmakers are deciding the price, then we're already wrong.
Each person perceives the subjective value of their own life differently. There shouldn't be a unified "value of life". Each person should do their own cost-benefit analysis and decide how much they're willing to risk their own safety. The incentives should be placed in such a way that each person's decision doesn't directly impact other people's lives though.
And maybe those that "want lower estimates" don't really want lower estimates, but they don't want any specific estimate to be shoven down their throats. In the same vein that those that want less safety regulations don't really "want more accidents", they want more agency and freedom of choice so that each person can make their own decision depending on their priorities.
So how would that work - I have covid, how do I determine if I should stay home? Should people I will potentially encounter throughout the day each individually decide how much they would pay me for staying home?
This is very silly; the worst case of a fetish for creating metrics to try to make something "objective" when the determination really needs to be made on the totality of human experience and moral reasoning.
The "ship pollution to poor countries" is a great example of this -- hey, we have this metric, and according to this metric, it's a net win. This is one path to doing monstrous things in the name of minimizing a metric, rather than actually making progress.
To convince yourself that the task of "value of human life" is futile, the following thought experiment suffices -- there are different penalties for deliberately causing the death of another (and distinction based on the quality and recency of those motives) vs. negligently causing a death vs. recklessly causing a death, in all justice systems invented by humans.
The article says that attempts to do this have "shown puzzlingly great variation across individuals and contexts", but then attempts to reduce it to new and interesting metrics instead. Uncertainty is the fact of life; this pursuit is meaningless except as a post-mortem examination of the impact of an action taken in ignorance of the metric. If the metric is involved in the decision making in any form it immediately becomes useless and counterproductive.
> To convince yourself that the task of "value of human life" is futile, the following thought experiment suffices -- there are different penalties for deliberately causing the death of another (and distinction based on the quality and recency of those motives) vs. negligently causing a death vs. recklessly causing a death, in all justice systems invented by humans.
How does that prove the futility of the exercise? It only proves that there must be more than one way to do it, and that the right method depends on context. There are metrics for use in criminal cases, civil cases, insurance, etc. It's not like we can fail to make these determinations -- failing to put a value on human life in each relevant context is akin to saying there is no value to human life at all.
> failing to put a value on human life in each relevant context is akin to saying there is no value to human life at all.
This is the crux of the problem with metric-based examination of nearly everything except physics.
It is not nearly akin to that. If anything, saying that you can have a metric that measures the value of human life is much more akin to saying there is no value to human life at all -- all of life is "haggling over the price" as it were.
Saying that it cannot be measured is just making an epistemological declaration of lack of knowledge. Not of the general form of "well, we just have to find the right variable and plug them in to the right formula" but of the form of "this is not well-suited to being reduced to a metric, as any such reduction will mask aspects of its value that are obvious but not easily measured".
To be clear on the dangers (as if the McKinsey-fication of "dump the pollution in pooristan" wasn't enough) the problem is that given an imperfect approximation to some theoretical perfect metric, we have absolutely no way of determining how imperfect it is. There's nothing to measure it against except other potential metrics. There's absolutely no concept of ground truth -- the universe will not correct you (as it will if you measure mass incorrectly when doing physics).
There's a lot tied up in here, so I'm going to attempt to untangle.
First, I don't think there is a universal model that we can apply to every possible case. Budgeting medical care, for example, will depend pretty heavily on funding cycles, and a surplus at the end of the quarter raises the question of "could this have been better allocated to serve patients' medical needs during the quarter?"
Off the top of my head... I'm thinking that something like a lifetime maximum with some sort of payback mechanism may be the answer. E.g., you get $2M of medical care over the course of your life, but when the medical system has a surplus, those numbers are adjusted.
Liability, on the other hand, is a whole 'nother ball game.
A manufacturer that deliberately chooses not to implement a safety feature because "the lawsuits will be cheaper" is making both a moral and an economic choice, and that's something that needs to be addressed via both torts and corporate law. Broadly speaking, I think that putting the customer -- rather than the shareholder -- back in the economic drivers' seat would do a lot to remedy this.
It's too bad there's no equivalent to "one cuts one chooses" for financial modelling. For instance, if we lived in a transhuman universe and the sick old person could trade consciousnesses with the healthy young economist simply by paying the difference in their QALY values (maybe funding sources would even enable the LDO, the leveraged die out?), QALY numbers might be different.
It discusses various ways that the value of a life have been calculated and how it has been used to determine whether product labels are worth changing to save lives. They suppose that the shutdown could have saved $10 trillion worth of lives.
> They suppose that the shutdown could have saved $10 trillion worth of lives.
I suppose it could have cost $10 trillion worth of lives. Not sure what the calculus of suppose is exactly, once I have figured that out I can take this to the bank.
> Infinite – Pay any price for any chance to save any human life.
> [...]
Say you take this position, that the value of a human life is infinite. Then you come to a situation where you have to make a trade-off: some must die in order for others to live. Something of a Sophie's Choice, though perhaps less dramatic. Now what?
This does happen! When it happens it's not necessarily obvious, but it happens. TFA gives an example.
Putting an infinite value on human life isn't practical. Putting a finite value on human life seems crass, but is practical. We just have to be a bit crass then -- what else can we do?
When in life outside of imaginary scenarios can this calculus actually be done? The entire idea here is ridiculous and relies on a predictive ability that nobody actually has.
It's like the trolley problem -- the reality of that problem is somewhere between "how the fuck should I know what this lever does, I'm not a trolley engineer" and "what if the people move".
If one person, Bob, seems likely to you, a police officer, to murder another person, Alice, then you don't say "well, Bob has greater expectations of higher lifetime earnings and Alice is too risk-averse". The idea that you can look at this outside of the direct context by examining a metric is ridiculous on the face of it.
All the time. People have to decide how much to spend on their own healthcare, for example -- in some cases it's governments that have to, in others it's individuals and/or their families. Courts have to decide how to award damages in wrongful death cases. Juries and courts have to decide how to sentence in homicide cases.
> The second view, where we put a specific dollar value on each life, has long been shunned by officials, who deny they do any such thing, even though they in effect do.
I don't know how much they deny it; what fascinates me is that different departments of the US government assign different dollar values to a life (e.g. when it is too expensive to mandate a change to the road building code).
so if human lives are priceless, then we are prioritizing it over other life on the planet. This may not have been a big deal until now that we have ~8 billion people. This worldview will eventually kill everything to preserve humanity, which will ironically kill humanity.
You aren't evil for saying human life is not infinitely more valuable than all other life. There's a point where the scale tips.
I enjoyed reading a book by the late Finnish conservationist Pentti Linkola, who said that to achieve healthy levels of biodiversity that the value of a species should become less valuable as it becomes overpopulated. (e.g. house cats are not as important as tigers right now, given tiger endangerment). It applies to humans.
Readit News
You can drive the suggested analysis here as far as you want, but ultimately, you will lift your head and realize it's all useless for actual policy making wrt COVID-19.
The ultimate answer is that there is no ultimate answer. Abstract mathematical modeling for these sorts of questions is therefore mostly useless. Useful answers come from going in and dealing with lots of messy particularities in a very messy reality.
I like to think that this problem with quantitative rationality, where you build a model about one thing and get a very precise answer, but then realize your answer is really just another question, is exactly what's being alluded to in "42".
(e: moved this sentence to the bottom of my comment because it seems to be causing some distracting hostility: "This sort of thought process is exactly how you go about convincing a wise person to just give up on the whole hard core rationality project. It's an unintentional strawman.")
All I see here is "The value of human life is incalculable, so just give up."
Am I wrong? What am I missing?
We don't have infinite resources or infinite capacity to distribute them. That means we need to make hard decisions about things like medical care. That is reality. If I get some kind of crazy-aggressive brain cancer that costs a billion dollars to treat, I fully expect to die.
The best way to make those decisions is to have some kind of model that can (a) be readily understood; and (b) helps us make those decisions fairly and with as little unintentional bias as possible.
There will always be bias -- it's part of the human condition -- but if we can be transparent about our biases and willing to discuss them, we open the door to improving our decision-making systems later on.
With that in mind, what would you suggest as a model for determining, say, how we budget for healthcare?
My solution is that the solution probably isn't useful so don't bother. Seriously. The game of assigning quantitative value to lives is almost never going to help you make better decisions.
> All I see here is "The value of human life is incalculable, so just give up."
Nope. It's exactly not that! I don't think it's incalculable. You can definitely choose a way to calculate it. Precisely, even. But it's going to be some complex function. And you're going to plug that function into another model. And when you do that, one of two things will happen:
1. you'll ground out in a deontology because you choose extreme values, or
2. the uncertainty in the other parts of the model will dominate.
In COVID-19, #2 is what happens.
> With that in mind, what would you suggest as a model for determining, say, how we budget for healthcare?
I suggest that any reasonable model would look more like an entire industry of mathematicians thinking about edge cases than what we would recognize as a single mathematical model. And even then the answer would probably suck. And definitely nowhere would you find it useful to have a quantitative assignment of lives vs. money, because anywhere you try to use that is going to have a ton of uncertainty that dominates the useful range of possible values. Which pretty much comports with my original comment that "there is no universal answer". I'm not an actuary or policy analyst, though.
It's easier for us as a species to say what a star billions of light-years away has been "doing" since its "inception" and the most probable way in which it will "die" than to say what the 5-year old kid next doors is going to do in a year or two, heck, what he's going to do tomorrow.
[1] Well, ideally. The US response to covid was pretty much [2]. If it wasn't clear, [2] was a rhetorical question and you should not do that.
42? I'm genuinely not sure what number you're asking for.
> Eventually somebody[1] has to make a high-level decision on which strategy to pursue.
Yes, of course, but that decision making doesn't require a quantitative model assigning value to life (or not). You can make a lot of very well-informed decisions without ever thinking about that particular aspect of optimization theory with any particularly high level of detail. "Aggressively avoid protracted exponential growth" doesn't require knowing anything about the value of life relative to other things.
Mathematical modeling can be useful for responding to covid-19. However, mathematical modeling for this question is not useful, because your answer gets plugged into yet another model where there's too much uncertainty for your answer to the have any quantitatively justifiable impact on decision making.
To give a toy example, spending a bunch of time figuring out the value of $c$ in $f(x) = x^n + c$ is a waste of time if you don't know much about the value of $n$.
Deleted Comment
References? COVID-19's behavior has been quite simple so far. Exponential growth with variations. It's more predictable than the weather next week, overall.
The American curve and European curve are somewhat different, yes but the differences can be explained by the total fucked-up the American virus control efforts have been.
Oh, and the American failures were accompanied by complaints about uncertainty and how much virus control efforts would cost. Now, we will find out how much a lack of control is going to cost.
What do you mean with behavior? People are still discussing how it spreads.
The latest news is it travels as aerosol and that in badly ventilated rooms keeping 1.5m distance is not going to help much. And also that the amount of virus particles you inhale determines how sick you get.
So it is only predictable when we know how it is growing exponential.
The still ongoing discussion about how it spreads creates uncertainty.
There is far less uncertainty around COVID-19 now than there was at the beginning of the year. It is possible to see which countries are successfully tracking and tracing to eliminate, which are entering second waves and need to mitigate, and which haven't even finished their first.
https://www.lemonde.fr/les-decodeurs/article/2020/03/27/coro...
For example they might claim in the early stages of an epidemic that we actually just have linear case growth. It is then up to the modeller to try to justify why that isn't the case, which in limited-information regimes can be near impossible in any rigorous sense, for the reasons you describe.
However, assumptions are important, and showing decision makers what might happen given certain assumptions (and providing a justification for those assumptions), which is part of what a model does, can cause them to reassess their current thinking, and it is for this reason that these kinds of models can remain useful, even if it is not possible to reasonably distinguish between some set of models given available data.
Returning to the epidemic example, justifying why exponential rather than linear growth is likely in the early stages of spread is something that can be done on an intuitive level, and explaining this to policy makers, and showing them the range of possible outcomes that are consistent with current (at the time limited) knowledge, can be extremely powerful.
An alternative way to explain the meaning of “large uncertainties” is that you can fudge input numbers slightly to get whatever output decision you want, so whoever is making the policy decision is likely to keep fudging numbers till the outcome “feels right”. At that point, the modeling is just dangerously misleading pseudo-science.
An intuitive approximation of the decision based on time averages is to “play it safe” in the presence of fat negative tails. “How safe tho?” is a very hard question to answer because of the uncertainties involved (as parent post explained).
If lawmakers are deciding the price, then we're already wrong.
Each person perceives the subjective value of their own life differently. There shouldn't be a unified "value of life". Each person should do their own cost-benefit analysis and decide how much they're willing to risk their own safety. The incentives should be placed in such a way that each person's decision doesn't directly impact other people's lives though.
And maybe those that "want lower estimates" don't really want lower estimates, but they don't want any specific estimate to be shoven down their throats. In the same vein that those that want less safety regulations don't really "want more accidents", they want more agency and freedom of choice so that each person can make their own decision depending on their priorities.
The "ship pollution to poor countries" is a great example of this -- hey, we have this metric, and according to this metric, it's a net win. This is one path to doing monstrous things in the name of minimizing a metric, rather than actually making progress.
To convince yourself that the task of "value of human life" is futile, the following thought experiment suffices -- there are different penalties for deliberately causing the death of another (and distinction based on the quality and recency of those motives) vs. negligently causing a death vs. recklessly causing a death, in all justice systems invented by humans.
The article says that attempts to do this have "shown puzzlingly great variation across individuals and contexts", but then attempts to reduce it to new and interesting metrics instead. Uncertainty is the fact of life; this pursuit is meaningless except as a post-mortem examination of the impact of an action taken in ignorance of the metric. If the metric is involved in the decision making in any form it immediately becomes useless and counterproductive.
How does that prove the futility of the exercise? It only proves that there must be more than one way to do it, and that the right method depends on context. There are metrics for use in criminal cases, civil cases, insurance, etc. It's not like we can fail to make these determinations -- failing to put a value on human life in each relevant context is akin to saying there is no value to human life at all.
This is the crux of the problem with metric-based examination of nearly everything except physics.
It is not nearly akin to that. If anything, saying that you can have a metric that measures the value of human life is much more akin to saying there is no value to human life at all -- all of life is "haggling over the price" as it were.
Saying that it cannot be measured is just making an epistemological declaration of lack of knowledge. Not of the general form of "well, we just have to find the right variable and plug them in to the right formula" but of the form of "this is not well-suited to being reduced to a metric, as any such reduction will mask aspects of its value that are obvious but not easily measured".
To be clear on the dangers (as if the McKinsey-fication of "dump the pollution in pooristan" wasn't enough) the problem is that given an imperfect approximation to some theoretical perfect metric, we have absolutely no way of determining how imperfect it is. There's nothing to measure it against except other potential metrics. There's absolutely no concept of ground truth -- the universe will not correct you (as it will if you measure mass incorrectly when doing physics).
First, I don't think there is a universal model that we can apply to every possible case. Budgeting medical care, for example, will depend pretty heavily on funding cycles, and a surplus at the end of the quarter raises the question of "could this have been better allocated to serve patients' medical needs during the quarter?"
Off the top of my head... I'm thinking that something like a lifetime maximum with some sort of payback mechanism may be the answer. E.g., you get $2M of medical care over the course of your life, but when the medical system has a surplus, those numbers are adjusted.
Liability, on the other hand, is a whole 'nother ball game.
A manufacturer that deliberately chooses not to implement a safety feature because "the lawsuits will be cheaper" is making both a moral and an economic choice, and that's something that needs to be addressed via both torts and corporate law. Broadly speaking, I think that putting the customer -- rather than the shareholder -- back in the economic drivers' seat would do a lot to remedy this.
It discusses various ways that the value of a life have been calculated and how it has been used to determine whether product labels are worth changing to save lives. They suppose that the shutdown could have saved $10 trillion worth of lives.
I suppose it could have cost $10 trillion worth of lives. Not sure what the calculus of suppose is exactly, once I have figured that out I can take this to the bank.
> Infinite – Pay any price for any chance to save any human life.
> [...]
Say you take this position, that the value of a human life is infinite. Then you come to a situation where you have to make a trade-off: some must die in order for others to live. Something of a Sophie's Choice, though perhaps less dramatic. Now what?
This does happen! When it happens it's not necessarily obvious, but it happens. TFA gives an example.
Putting an infinite value on human life isn't practical. Putting a finite value on human life seems crass, but is practical. We just have to be a bit crass then -- what else can we do?
It's like the trolley problem -- the reality of that problem is somewhere between "how the fuck should I know what this lever does, I'm not a trolley engineer" and "what if the people move".
If one person, Bob, seems likely to you, a police officer, to murder another person, Alice, then you don't say "well, Bob has greater expectations of higher lifetime earnings and Alice is too risk-averse". The idea that you can look at this outside of the direct context by examining a metric is ridiculous on the face of it.
I don't know how much they deny it; what fascinates me is that different departments of the US government assign different dollar values to a life (e.g. when it is too expensive to mandate a change to the road building code).
Driving 10km/h slower to save lives? People could accumulate hundreds of man-years essentially not lived while driving very quickly.
You aren't evil for saying human life is not infinitely more valuable than all other life. There's a point where the scale tips.
I enjoyed reading a book by the late Finnish conservationist Pentti Linkola, who said that to achieve healthy levels of biodiversity that the value of a species should become less valuable as it becomes overpopulated. (e.g. house cats are not as important as tigers right now, given tiger endangerment). It applies to humans.