TLDR: Is it plausible that the world is only as good as its worst phenomenal moment? Should raising this bar be the aim of altruistic action rather than doing the most good in aggregate? If so, what does this mean for cause prioritization?

Disclaimers:

• I’m not particularly trying to argue for or against any point of view, but I’m rather just trying to put some intuitions to writing to see what others make of certain lines of thinking that I can’t 100% accept nor 100% set aside.
• I come into this post with an implicit assumption that preventing suffering is the primary moral aim (as opposed to things like preventing the absence of positive experiences). I feel like the general discussion around the importance of suffering has been extensively discussed by CRS and others. Rather than justifying this starting point, I wanted to focus on one area of concern  probably for an audience already convinced that reducing extreme suffering is the most important action. I recognize that the way this post looks at the world is singularly through a lens of the worst experience, disregarding views that hold individuals as having holistic, inherent value.
• I have not cited my sources super well since a lot of this is a jumble of a bunch of ideas have come across, but I have hyperlinked wherever possible. 

Summary

I’m writing this post to put some intuitions I have around aggregating experiences into words to clarify the end to which my altruistic actions should aim. To summarize, the problem I cannot figure out how to address is that the value/disvalue we assign to an aggregate of experiences is never actually realized by anyone (see this blog post for more on this). This sum simply exists in an impartial universe. This creates a problem for me when trying to compare suffering of different severity levels where it is unclear that the badness that is felt by one entity can ever be outweighed by some aggregate amount of less severe badness felt by any number of other entities. Following this train of thought, I am led to believe that to address the pain that is felt, I should focus my attention on improving the circumstances of the very worst of phenomenal moment in the universe as Joseph Mendola puts it (quoted in this CRS article). After all, the true terror and urgency of suffering is primarily bad because it is felt and secondarily because it exists as an aggregate sum. 

If anyone has any feedback on what this implies for cause prioritization or what flaws/benefits there are in this way of thinking, that would be helpful. Separately, if anything is written poorly or unclear, I would be happy to make adjustments there too. The rest of this post is just more elaboration on the concept introduced above, so it is not critical to respond to the problem. Feel free to skim or skip if pressed for time.

Background & Purpose of Post

After reading Magnus Vinding’s Suffering Focused Ethics and consulting several of the references in the book, I have become more uncertain about experience aggregation and whether events of different severity can even be compared. Unlike one of the initial charts I came across after being exposed to ideas around EA where we multiply scale and severity to get total impact (see here), I have started to believe that severity and scale cannot be compared in this way because the former is actually experienced while the latter exists as a descriptive sum in an impartial universe. When looking at neutral/positive experiences, it doesn’t feel too problematic to mix the two. However, when considering extreme suffering, it seems far more problematic because of how terrible extreme pain is to experience. Certain thought experiments involving disregarding the felt severe experience of even a single living being for an aggregate of less severe experiences experienced by multiple living beings are particularly disturbing (see this blog post for some, extensively discussed in Suffering Focused Ethics, and the torture vs dust specks thought experiment). Problems that affect many living beings are certainly all the more terrible, but the true terror of suffering is in the severity of what is felt. The former and latter are different dimensions of what makes a problem bad with the latter more salient in cases of extreme suffering. To illustrate, I will introduce a though experiment in the next section that illustrates why I think the differences in the badness of severity and scale are of two separate kinds.

The Experience Wiper

Let’s consider a 24-hour experiment where patients will be subject to moderately painful electric shocks and excruciating electric shocks. After each shock, the subject is fully cured, and the experience is wiped from memory such that the person feels like they just entered the experiment room. After the experiment, there will be no memory or physical damage. It will be like the subject time traveled 24 hours.

Now let’s consider two possible scenarios. In one scenario, the subject is moderately shocked 500 times (each shock being cured and wiped from memory before the next shock). In another scenario, the patient is severely shocked once (with one shock cured and wiped from memory). Which one is worse? While there may be more aggregate pain occurring across time in the first scenario, the pain that is actually felt is greater in the second scenario. The sum of 500 shocks occurs but is not experienced while the excruciating shock is actually experienced. If I were the subject, I would want to reduce the severity of my first-person experience rather than a discontinuous third person aggregation even though I leave the experiment unharmed in both scenarios. 

If we really stretch this example, the world actually doesn’t seem too different. While the physical & mental structure of all sentient beings is incredibly diverse, all these beings are all conscious experiencers of suffering. Individual humans, cows, pigs, fish, and possibly insects are all like the different episodes of experience in the experiment above. The sum of experiences across all these beings is not the item that is felt or that is terrible, but rather the terror of just a single individual experience is already as bad as it can get along the dimension of severity. Thinking about this same concept in another way, if I knew I would be reincarnated as every living being (analogous to the electric shock experiment), how would I want the world to be? Personally, I would want to raise the bar of the very worst I would have to experience rather than improve some aggregate or average well-being across beings since there is no master self to realize this aggregate. Whereas the value of the former will fully be realized in experience, the value of the latter will not be realized because every attribute except for consciousness changes every time I reincarnate. All injuries, memories, and individuality are wiped every time I reincarnate. In this sense, the person I think I am right now isn’t really reincarnating, but rather it is the general sense of awareness that is common across all sentient beings that is reincarnating. If I consider this general sense of awareness as myself and the source of all moral value, then I really am in some way all living beings. It is within this single sense of awareness that doesn't exist as a quantity but rather a quality that intrinsic badness occurs. Fundamentally, if we care about suffering for the simple reason that it feels bad, we should reduce it at this level: The level at which it is felt. I would be willing to do a lot to prevent the realization of that severe electric shock but would be less disposed to preventing the realization of many discontinuous minor shocks however much they occur because there is not as much felt badness intrinsically begging for it to stop.

What About Aggregation Across Time Within Living Beings?

If aggregating experience across multiple living beings has the aforementioned issue, can the same be said of aggregating experience within the same living being across time? Even if interpersonal aggregation has flaws, does intrapersonal escape these?

On first thought, aggregation across time seems like a reasonable approach to evaluating the badness of pain. If I consider being electrically shocked twice in a certain span of time versus being shocked once, it surely seems like the former is worse because there was another instance of pain. And clearly, experiencing less instances of pain is better than experiencing more instances of pain. Once again though, let us again consider the possibility that my memory is wiped and my body cured in between the shocks. Whether I am shocked once or twice, I will only remember one shock, but all else equal, there is more pain in the universe when I am shocked twice. It is probably true that two shocks with full memory is worse than the two shocks with partial memory because the first case involves some psychological continuity to aggregate over. However, how would the ranking change if the partial memory case consisted of 10 shocks instead of two? Is there a number of shocks with the memory wipe & physical cure in between shocks that could outweigh 2 shocks with full memory? What I am getting at here is that it may not be the fact that two shocks are administered that is bad but rather the fact that the first shock is encoded in memory and inflicts damage to the body before the second shock is administered. There is something about the fact that two shocks are experienced in psychological continuity that makes it worse than one shock rather than the mere existence of two shocks.

As a whole, I definitely think things become a bit more uncertain here, but one possibility to consider is that perhaps suffering does not occur across time but rather time (or our mental representation of what has occurred) is an input for suffering. Being shocked twice in a row may not be bad because it is two instances of a single pain, but it may actually be pain of a different quality (not a greater quantity). As this ReThink Priorities article discussing severity versus duration notes at one point, pain may be an ordinal rather than a cardinal phenomenon. Pains of long durations and high quantities within a single living being may be bad to the extent that they can influence the quality of experience as a whole (ex: they physically create a more fragile body or influence perception of future events), but the aggregation is not intrinsically what makes the multiple pains bad. The problem of whether many instances of less severe pain can outweigh a single instance of severe pain is less about comparing quantities across time and more about deciding whether the many instances of a less severe pain create an overall disvalue that outweighs the more severe pain in an ordinal manner. It is true that aggregation may be a proxy for deciding how bad a sum of experiences is, but this badness is dependent on the degree to which these quantities can be encoded in memory to affect future experience rather than from the sum of pain being intrinsically bad. It is true that this perspective may be tainted by scope bias where I can’t adequately evaluate many instances of pains, but I also don’t experience the disvalue of many instances of pains as a linear addition because past pains exist as a memory reinforcing future pains or physically as damage to the body. 

From the perspective of experience, there is just an abstract timeless sense of badness, which is a result of duration, stimulus intensity, quantity, emotion, and the myriad of factors sentient beings consciously process. These factors are only valuable to the extent that they are experienced and encoded rather than due to their objective values. To illustrate, I think this sort of thinking may be consistent with views that hold that the subjective time is more important for evaluating experiences than objective time (see subjective time explored here and here). For example, an unpleasant event like a 1 minute heat stimulus may feel more terrible for a being that experiences time faster than for one that experiences time slower. One way to explain this is that there is a greater quantity of experience in that 1 minute interval for the first being, so the sum of pain is greater. However, this could also be explained by the heat stimulus reaching a greater degree of terribleness within that interval for the first being than for the second being. The way time is experienced is the input for suffering rather than the objective duration of the stimulus. To summarize, I think experience of time may be considered an input for the overall badness of the pain that is felt rather than time being something to add a certain quantity of experiences over.

Overall, while this perspective attempts to object to intrapersonal experience aggregation, this is not perfect and opens up other problems. For example, without acknowledging the badness of an additional instance of pain for its own sake, we may have to be neutral between being electrically being shocked 10 times and being electrically shocked once but having the mental representation & physical damage of the first 9 shocks administered before this one shock. Assuming the 10th shock creates the worst overall quality of experience, these two scenarios would be identical in terms of the severity threshold they reach and would thus be evaluated the same by a worldview that aims to reduce the severity threshold of what is experienced. Another more mentally confusing aspect of this is whether I really stay true to looking at time within suffering rather than suffering within time. The way the example is phrased is as if the worst overall experience results at a moment in time potentially implying that I care about a suffering in a slice of time.

Implications Of This View & Other Thoughts

S-Risks + Proportion vs Quantity

One argument for S-Risks I have heard in Magnus Vinding’s Suffering Focused Ethics and other material on the Center for Reducing Suffering’s (CRS) website is that worst case outcomes make up the majority of expected suffering in the universe. Essentially, it may be better to try to reduce the most suffering in expectation rather than the most in proportion. For example, reducing 50% of the suffering in the worst possible future outcomes may amount to a greater overall reduction of suffering than reducing 100% of the suffering in a default scenario (perhaps saving billions of more beings in the former than the latter). This is one argument made against certain causes like David Pearce’s Abolitionist Project that seeks to use biotechnology to eliminate suffering in the natural world. Eliminating 100% of suffering (particularly extreme suffering) using this method sounds like an excellent achievement, but if we want to make the best expected value decisions, it may be better to devote resources to preventing the worst possible outcomes from being realized.

From the perspective of reducing suffering in aggregate terms, I think these reasons certainly make sense, but from the perspective of reducing extreme suffering along its most terrible dimension (the fact that it has to be experienced by even a single living entity), it may actually make more sense to try to eradicate it 100% or at least eradicate working upwards from the worst phenomenal moment. Unfortunately though, this implication may be quite implausible because it implies that certain gambles can rob all resources. For example, imagine there is a burning building. We can take one course of action that saves 99 people from the fire with 100% probability. We can take another course of action that saves all 100 people in the building with 1% probability. Theoretically, if we ignore any further implications from lives saved and only focus on the suffering at hand(burning alive), then the first course of action does not reduce the severity threshold of experience whereas the second course of action actually has a chance of reducing the severity threshold. Whether 100 people are in the building or 1 person is in the building, the pain that is felt is the same (assuming each mind experiences the same pain from the fire). Experience is one even if experiencers are many.

Seeing as this would be the result of valuing severity strictly over scale, it seems like folks who have thought about this a lot tend to adopt hybrid views where extreme suffering is lexically or at least exponentially more terrible that milder suffering but that the goal is still to minimize the quantity of extreme suffering. See value lexicality discussed extensively on the CRS’s website for how we can attempt to aggregate while giving priority to the worst.

Animal Related Causes

As an animal sympathetic person, I find it a bit unnerving that this view may imply that animal related causes are less important than before because a huge part of what makes them so salient is the massive numbers of individuals involved. Still, even apart from the numbers, the pain farmed & wild animals have to endure is incredibly severe but perhaps not as severe as intentional sadistic or violent actions humans inflict upon each other.

Within farmed animal cause areas, this view may imply that is less important to favor incremental progress in improving the lives of many animals over actions that can displace animal agriculture all together. For example, whether or not reform XYZ is made in a facility, there will potentially be at least one freak accident involving pure agony for at least one chicken given the huge numbers that are processed each year. The severity of what is experienced is potentially left unchanged with incremental progress just because the system exists. If we are trying to reduce the pain that is felt rather than an aggregate of the pain that is, it may be more worthwhile to try to pursue strategies in alternative protein or more disruptive changes over incremental changes.

Research Into Extreme Suffering

Another implication of this is regarding research into extreme suffering itself. If understanding terrible mind space in order to prevent it from happening in the future requires even one entity experiencing this terrible mind space, the worst dimension of pain (what is felt) has already been reached.

Philosophy of Time

One huge problem with this perspective is that it may depend on whether we take the past as having existence. If so, some of the most terrible experiences have already been realized with no hope of them ever being reversed. From the perspective of reducing the severity threshold of experience, is it even possible to improve the world from the dimension of pain that is felt? 

It seems like some form of aggregation is needed where we accept the value of preventing additional instances of terrible experiences. Perhaps, we can say that the best thing to do is reduce the severity threshold from here on out. Within that view, helping a greater number of entities not realize a certain type of pain is valuable both because it brings us closer to a complete elimination of that experience from existence and because we take scale to matter on its own accord. 

Still, I think considering this implication may be the biggest blow to such a view since so much of altruism focuses on preventing bad things which have already been realized from happening again. Perhaps, the main cause area with this single focus would be to ensure the severity level of experience does not exceed what it already has and prevent specific types of malevolent, intentionally cruel future actors from enacting their wishes. However, this is quite strange since it is letting what has already happened affect what we focus on in the future, so I don't think this quite works.

Many World Interpretation of Quantum Mechanics

I’m not going to pretend like I really understand this, but from what I can gather, this interpretation implies that different versions of the worlds literally do exist. If that is the case, then outcomes where terrible suffering is experienced and where it is not are both realized. In that sense, we can never wipe out extreme pain from the dimension of it being felt. Perhaps, we can simply just reduce the number of individuals who suffer from terrible experiences, but if there is infinite value/disvalue in the universe, we don’t ultimately change anything in aggregate either.

Suffering Subroutines

Brian Tomasik writes about suffering sub-routines as being sub-agents within larger agents that can suffer. To my understanding, just as I am an agent that experiences suffering, the different mental processes that make up myself, and the physical processes that make up the universe may have some degree of suffering. Theoretically, I think the viewpoint of reducing the severity threshold by identifying the worst phenomenal moment whether in a sub-agent or consolidated agent still holds, but the empirical challenge is much greater.

Conclusion

Thank you to all the thinkers who have thought about these subjects before me and given me the base from which to engage in my own reflection on what ought to be done. This line of thought may be a dead end, but I am still curious to see if there are any ways to make sense of these intuitions. Once again, my main goal in writing this is just to help put some intuitions into words to help make better altruistic decisions going forward rather than advocate a view. Ultimately, on the empirical front, I have pretty standard decisions to make around earning to give versus direct work, and I hope to sort out any theoretical jumble before making the more impactful life decisions ahead of me. If anyone has any insights into this, that would be much appreciated. For now, I think I’m leaning to just sticking with a general extreme suffering focus without following this line of thought to the maximum implications.