I'm trying to get the decision to avoid the vaccine bumped up to people's system 2 to see what kind of justification they come up with. — Isaac

That's generally what we do around here, right.

This is just a big mess for philosophy in general: on the one hand we want to talk as if everyone is in System 2 mode, but we're regularly dependent on data from people's System-1-driven behavior. (What philosophers are accustomed to call our "intuitions" -- without those there's no Gettier problem, not much to talk about in ethics, linguistic evidence is worthless, etc.) That's fine-ish, but it makes collecting the System 1 data awfully confusing, or, rather, it makes it hard for the one providing the data to know if you want the gut reaction or the rationalization, and obviously most people prefer to present their rationalizations to the world. (New posters here who present only their gut instincts about everything and no rationalizations are shunned and eventually banned.)

This is just a big mess for philosophy in general: on the one hand we want to talk as if everyone is in System 2 mode, but we're regularly dependent on data from people's System-1-driven behavior. (What philosophers are accustomed to call our "intuitions" -- without those there's no Gettier problem, not much to talk about in ethics, linguistic evidence is worthless, etc.) That's fine-ish, but it makes collecting the System 1 data awfully confusing, or, rather, it makes it hard for the one providing the data to know if you want the gut reaction or the rationalization, and obviously most people prefer to present their rationalizations to the world. — Srap Tasmaner

Yes, that's exactly it. And the intricacies that go along with it.

Not all gut instincts can be said out loud, some social situations demand only the rationalisations, in others rationalisation sounds cold and overly pedantic. Philosophy is, afterall, a social activity. It has it's rules too. Like how a system 1 quip can trump a system 2 castle-in-the-air (see "trees don't really exist" - "well, duh, there's one outside my window"), but played wrong the same move comes off sounding naïve and unsubtle. I get it when reading some of the more complicated philosophy, part of me's carefully putting piece to piece constructing the jigsaw until, occasionally, system 1 says "but this is all bullshit, isn't it?"

Here I think it's very much about the social narratives though. I'm guessing, but my gut instinct is that people don't have much of a gut instinct about vaccines as individuals, rather there are a number of social narratives available; the diligent follower of Science™, the see-through-it-all conspiracist, the 'ass-kicking' honor-and-duty soldier, the 'actually-I-think-you'll-find' academic...

Despite titling them facetiously, I'm not suggesting any are somehow insincere (I'm one of them), it's what we do, we use narratives to make sense of the otherwise borderline chaos and just haven't the bandwidth to be constructing individually tailored versions, so we pick them off the shelf, tweak them a bit maybe and then elbow everything into one. I like to see how other people do it, but unless people are confronted with a conflicting narrative they don't do the work, so all you get is the unadulterated story, the off-the-shelf version. Not so interesting.

The one thing I'd add is that since we know system 2 is the training ground for system 1 -- note -- there's hope that we can attend to our system 1 driven "instinctual" responses, and then reconstruct the system 2 analyses that went into them.

But it doesn't work like that. What your analysis of your system 1 instincts produces is a new product; there's no simple way recover the original process. What's more, system 1 does its own thing with your earlier system 2 efforts: it's purpose is to produce cheap, readily accessible summaries of the results of all that system 2 work. It is by design unfaithful to them. It's a lossy format.

So reconstruction isn't really in the offing, and all we can do is let system 2 do its job as it's currently constituted and in the current circumstances. (All the overcoming-bias types are happy if they can just get people to recognize when they've left unexamined a response it behooves them to examine.)


Note: At least that's my understanding. When first driving a car you have to think about every little thing and make lots of conscious decisions, but eventually it becomes second nature and you do it all without thinking

@Isaac Some time ago i requested of you a precis of your argument or position such as it is, which you, then, instead of providing two or three well-crafted sentences, answered by referring me to the 200 pages of this thread. The substance of that such that I find you a weasel in a box. In a box, to be sure, but not-so-easy to get hands on, possessing something of the agile slipperiness of the creature named.

And I don't get it. To me it's simple. There's a guy over there shooting at you, a guy over here offering you a bulletproof vest, and you full of jabber. And if it were that simple, that would be an end of it. But it isn't. Even the mere possibility of your being "wounded" - infected - is a communal issue, a concern of and for the community, your community. Or are you such a hermit in your thinking you no longer cognize community. Your rights v. theirs? Don't make a category error. They are individuals whose rights are no less than yours. Which means that as a community, theirs can trump yours.

Yes, I have a lot of respect for Kahneman's work, but it's not always easy to translate my own models through his. I think in terms of predictive models, of increasingly higher orders of generality which then feed back to models of lower order of generality, so for me there's two hierarchy's going on - the prior/update (assume your priors until they are overwhelmed by evidence the contrary, then update them) but also the general/specific relationship (create or update priors based on which would best support the priors of the model they form one of the data points for).

So with your car driving example, there'd be assumptions (priors) about where you should put your feet or hands next which you wouldn't question unless you were getting some really weird feedback (sort of like system 1 taking over), but your priors about where to put your hands and feet are not just modulated by previous system 2 work at the same level (learning to drive), they're modulated by higher order models - what sort of thing a car is, what it's purpose is etc. and the priors for those are modulated by even higher order models - what sort of thing is a car likely to be given what sort of thing the world is?

I've skipped a lot of steps in between for brevity, but you get the picture (hopefully). So social narratives form these higher order models which means that although they themselves cannot simply be swapped out (the have the prior vs update bias), if they are swapped out, they do put a backwards acting pressure on the priors for lower order models.

If that makes any sense at all?

I don't get it. — tim wood

Yet somehow do get it sufficiently to warrant an insult.

To me it's simple. — tim wood

Well then we can resolve our issue right away. We're facing a new form of virus which we're frantically working out the effects of in terms of both short and long-term immunology and physiological impact, it's affecting some of the most advanced and complex socio-economic units the world has ever seen at an unprecedented level of global interconnectedness. The medication is of a never before used technology, created by an integrated network of funding, private investment and government incentive. We don't know where this thing came from, it's mutating faster then we can update our medicines for it and in unpredictable directions. The models used to work out the effects of various strategies had to be run across several universities becasue no one university had sufficient computing power to include all the variables...

...and you think it's simple.


The world is not a Disney film, it's not divided in heroes and villains (conveniently colour-coded) there's no magic weapon to kill the evil dragon and people rarely live happily ever after. Come back to me when you've come to terms with that and we can have a grown-up conversation about some of the complexities.

Interesting but pessimist and depressive news we should to check out about the new variant...
What is the Mu variant of COVID-19 the WHO is now monitoring?

I lost the contract... — Isaac

But gained my respect :sparkle:

But gained my respect :sparkle: — ArguingWAristotleTiff

That's very kind of you to say.

If that makes any sense at all? — Isaac

Absolutely. Introducing hierarchy is very sound idea.

I have lots of thoughts which I am, through sheer force of will and adroit use of the "select all" and "delete" commands, not just vomiting all over your screen.

I think in terms of predictive models, of increasingly higher orders of generality which then feed back to models of lower order of generality, so for me there's two hierarchy's going on - the prior/update (assume your priors until they are overwhelmed by evidence the contrary, then update them) but also the general/specific relationship (create or update priors based on which would best support the priors of the model they form one of the data points for). — Isaac

Is this to say that as you move up a level in the hierarchy, you have a model that generates predictions about what models directly below it will be successful? Is there a rock-bottom where the models generate predictions about experience? (Trying to capture with "experience" just that we're talking about data that is not composed of models succeeding or failing, whatever it is composed of.) And then everything above is models of models?

I have lots of thoughts which I am, through sheer force of will and adroit use of the "select all" and "delete" commands, not just vomiting all over your screen. — Srap Tasmaner

Ha! I appreciate the editing, I have to do the same sometimes.

Is this to say that as you move up a level in the hierarchy, you have a model that generates predictions about what models directly below it will be successful? — Srap Tasmaner

More about what data they'll forward. The aim is always to save energy, and surprise takes energy, analysis takes energy, filtering out noise takes energy, so there's an incentive to pre-filter the noisy, potentially surprising data from models lower down (this is actually observable in neural activity, we're definitely stepping on the threshold of science proper here, even if we're not quite going in!). So the higher models are sending back a message like "I'm expecting a table, don't bother sending me any data that doesn't conform to the idea of a table".

Or, much more controversially, "I'm expecting this person I'm talking to to say things like my model of a hero/villain (delete as appropriate), don't bother sending me any interpretations of sentences that don't conform to that idea"

s there a rock-bottom where the models generate predictions about experience? (Trying to capture with "experience" just that we're talking about data that is not composed of models succeeding or failing, whatever it is composed of.) — Srap Tasmaner

Yes, (if i understand you correctly), sensation and interoception. The edge of our Markov Blanket. but we couldn't really call them experiences, it's more like 'edge', 'light', 'pressure', 'heart rate'... dozens of hierarchical levels later...'table', 'cup'....'my desk', working', 'the philosophy forum'... etc. But it's important to remember that this is an interactive process. at each stage there's a reaching out into the environment, we actively try to make it conform to our models of it too (again, it's just less energy that way). we're like terrible scientists, p-hacking our experiments all the time and concluding "oh yes, the data proves my model".

All of this regarding perception is fairly well established, the leading model at the moment. My work is (was) much more speculative relating it to much higher hierarchical levels of social narratives and how they filter and suppress signals from core beliefs, language interpretation, ethical judgement etc.

And then everything above is models of models? — Srap Tasmaner

Yes, that's right. An issue which usually stops people in their tracks, as if there was some sort of hideous circularity in that, but yes - models of models of models...

Gosh, we're miles off topic. Sorry.

At 207 pages long the topic finds it's way. I appreciate your sharing.

Gosh, we're miles off topic. Sorry. — Isaac

I still have the case at hand in mind and will be coming back to it.

(Most threads end up being about one of a few evergreen topics, no matter how they start.)

Was that it? Two hundred pages of your saying that gosh, Covid isn't a Disney film and some of the issues around Covid are not-so-simple? And here I thought the issue was your being against vaccination. I guess I misunderstood.

I still have the case at hand in mind and will be coming back to it. — Srap Tasmaner

Good. I'm happy talking about psychology in general though (too happy perhaps). Off to the pub now though, so it'll be tomorrow.

I thought the issue was your being against vaccination. I guess I misunderstood. — tim wood

Hard to misunderstand

The Covid vaccination programme is unquestionably an excellent public health initiative, as most vaccination programmes are — Isaac

Which bit of that confuses you?

Does that mean you're for it? By for it, I mean making the taking of it mandatory for many/most people? Keeping in mind that in the US at least, for example, vaccinations are required for students to attend schools.

<oops>

Does that mean you're for it? By for it, I mean making the taking of it mandatory for many/most people? — tim wood

So still thinking in Disney terms then? Either Thor's magic hammer or the poisoned apple?

So the higher models are sending back a message like "I'm expecting a table, don't bother sending me any data that doesn't conform to the idea of a table".

Or, much more controversially, "I'm expecting this person I'm talking to to say things like my model of a hero/villain (delete as appropriate), don't bother sending me any interpretations of sentences that don't conform to that idea" — Isaac

What follows is a little disjointed because I kept pruning away the joints:

It sounds like filtering is not something done by a subsystem that has that purpose, some bit of business we could properly call a "filter"; rather it's a way of describing how a model at one level constrains the models below it.

One typical feature of hierarchy is that there's no transparency across levels. That is, middle managers have some authority over their domain, and within broad guidelines are only told what to get done, not how to get it done. Upper management needn't even know much about the various roles and responsibilities of employees several layers below them. "I just need to know quickly if that's a knife and I don't want to hear any minutiae about visual processing."

When there is failure, i.e., surprise, it would, by definition, be contained at the lowest level it can be (unless there's some special provision made).

It's a wonder that we can communicate at all because a system like this is designed not to acknowledge novelty unless it absolutely has to, despite the obvious facts that everyone we speak with is unique and nearly every sentence we hear has never been spoken before. (This needs a lot more thought.)

Now some remarks about the sorts of discussions had here:

Pigeonholing is common, and it's just surprise containment. You attribute to another a view you are already familiar with instead of grappling with novelty.

It often seems to me that no one here really believes in disagreement, despite the above: it is incredibly common to see people in effect take the position that if another disagrees with me it must be because they don't really understand my position. (So I'll explain it again.) This is actually pretty uncomfortable, and seems like the sort of irritant the system is designed to minimize; when we speak candidly, we speak assuming that we will be understood, so to remain in a position of continuing to believe we are not understood is odd. But it can make sense from the production side: we don't need to say anything new, and can just repeat ourselves.

Both issues call attention to the likelihood that at least some of such a system is built around satisficing rather than optimizing. Even if the models throughout the hierarchy are mostly Bayesian, so that in effect you're continually running parallel best-first searches, there have to be some operational shortcuts to safeguard efficiency: a single model running too long before reporting back a result has to count as a failure; if you run multiple models at once, the first one back with a result probably wins. (Pinker talks in Words and Rules about some evidence for such races.) Satisficing is by definition good enough, and by design cheaper than holding out for an optimal result, but it's still a shortcut.

Satisficing has obvious negative consequences in discussions such as ours: people make the first criticism that comes to mind, without reflecting that a problem that obvious would likely have been noticed by the speaker as well (see @Nagase's exasperated dispelling of the myth that Logical Positivism was founded upon an obvious logical mistake); people resort to, shall we say, "extra-logical" strategies (deflection, ad hominem, obfuscation, and the rest) -- that might seem not to meet a presumed constraint of reasoned response, but if reasoned response is hard to come up with, some other kind of response will have to do; loss of perspective is common, opponents focusing on small issues while the main point of contention recedes into the background.

*

Bah. I had expected to get back to the coronavirus debate in this one, but this already way too much. I don't feel like I've advanced the discussion, but at least you can tell me if I'm in the neighborhood of your thinking.

It sounds like filtering is not something done by a subsystem that has that purpose, some bit of business we could properly call a "filter"; rather it's a way of describing how a model at one level constrains the models below it. — Srap Tasmaner

Yes, that's right

It's a wonder that we can communicate at all because a system like this is designed not to acknowledge novelty unless it absolutely has to, despite the obvious facts that everyone we speak with is unique and nearly every sentence we hear has never been spoken before. — Srap Tasmaner

Even further off-topic, but there's a whole slew of theories around language which tie in to this, all to do with the idea that we don't speak for communication at all (I'm exaggerating for rhetorical effect, of course - undermined by these parentheses telling you that, nevermind). The idea is that we communicate in whole sentences, the meanings of which are very broad and not very diverse (lots of sentences mean the same thing), but the detail of which is more like a form of art, personalisation to help with individual identity, group membership tokens etc. So it may not be quite so surprising afterall. We don't really have that much to say to each other in terms of meaning, but we do have a lot of feather-preening to do.

Pigeonholing is common, and it's just surprise containment. You attribute to another a view you are already familiar with instead of grappling with novelty. — Srap Tasmaner

True, but with the caveat that we're doing it to ourselves too. It's not that we're all unique snowflakes really and outsiders keep putting us in boxes. We put ourselves in boxes too, we interpret things other people say as if they were caricatures, but we also take what they say into our belief systems as if we were caricatures too. When the villain speaks it's not only that everyone interprets what he says as though he were the villain, it's that they interpret everything he says as though they were the hero/plucky sidekick/victim...

when we speak candidly, we speak assuming that we will be understood, so to remain in a position of continuing to believe we are not understood is odd. — Srap Tasmaner

See the paragraph above. It may not be that odd at all...in some interpretations of communication acts.

there have to be some operational shortcuts to safeguard efficiency: a single model running too long before reporting back a result has to count as a failure; if you run multiple models at once, the first one back with a result probably wins. — Srap Tasmaner

Yes. One surprising outcome of neuroscience getting involved in all this was the odd effect simple proximity (in terms of location in the brain) had on which models got their messages accepted most often. Like you say, the ones which got there first.

Satisficing is by definition good enough, and by design cheaper than holding out for an optimal result, but it's still a shortcut. — Srap Tasmaner

Well, to an extent, yes, but a shortcut to what? Are you assuming we want the truth? There doesn't seem to be much need for it. Least surprise in the long run perhaps... All told, our main aim seem to be much more short-term - personal identity, immediate threats, group dynamics... all well above learning the truth about the world as goals. Even for scientists themselves, truth is merely instrumental, having it can be used to further those goals within the narrative they're playing out. Like having the magic spear with which the hero slays the dragon.

Satisficing has obvious negative consequences in discussions such as ours: people make the first criticism that comes to mind, without reflecting that a problem that obvious would likely have been noticed by the speaker as well (see Nagase's exasperated dispelling of the myth that Logical Positivism was founded upon an obvious logical mistake) — Srap Tasmaner

So true. I read @Nagase's piece and enjoyed it, though the responses show already the erosion of such stellar arguments in the acid of a social narrative in which 'everybody knows' the logical positivists undermined their own project - they're famous for it. We just get too disoriented when they're removed from the story like that. You can't just kill off the main character in act 3. Because the accusation in the first place was never a reasoned assessment of the facts, it was the casting of a role in a play. As Mark Twain (?) said you can't use reason to disabuse someone of a position that wasn't derived using reason in the first place - or something like that.

at least you can tell me if I'm in the neighborhood of your thinking. — Srap Tasmaner

Yes, on the whole, very much so. I think the only difference perhaps is that I don't see discussions like this (the coronavirus one, not ours now) as a means to convince people of, nor collaboratively derive any kind of position, so what you might see as flaws, I see as features.

Like having the magic spear with which the hero slays the dragon. — Isaac

Heh. In the context of this paragraph and your last messages to @tim wood, this is an amusing substitution: in Disney's Sleeping Beauty, the weapon with which Prince Phillip slays Maleficent (in dragon form) is the Sword of -- wait for iiiiit -- Truth.

Are you assuming we want the truth? There doesn't seem to be much need for it. Least surprise in the long run perhaps... — Isaac

Actually, yes, that was what I was thinking. Quite short-term gains in efficiency, or gains within a department, could be overall inefficient, or in the long-term inefficient. It's a danger hierarchies are prone to by nature. Examples from the business world are endless.

As for truth, sigh. I get the argument, both because I read Nietzsche a million years ago and because my ride home from work the other night was spoiled by an interview on the radio with Donald Hoffman.

My working assumption goes something like this, speaking very loosely: evolution selects for an organism to have certain capacities that meet a need, but that doesn't mean those capacities are limited to meeting that need. We didn't evolve to be able to play baseball, but we do. I even have a pet theory that language is an accident, that we got an upgrade on our signaling ability that is far greater than any species could ever need. Satisficing can also give you something better than you asked for.

And so it could be with our ability to apprehend the truth: may not be what we were "designed" to do, but that doesn't mean we can't. It does mean it's worthwhile understanding what limitations may be built into our capacities -- there's even a long-running debate about whether throwing overhand, as in baseball, is unnatural and inherently injury-producing! -- and god knows it means being aware of what your subsystems will do if left to their own devices. ("System 1 is a machine for jumping to conclusions," as Kahnemann says.) We don't know that we can apprehend the truth, but we also don't know that we can't, and we can make the attempt. (I want to say, "can make the attempt and see", but -- trouble. We evidently can play baseball, so that's something. <Insert joke here about your favorite team's effort last night.>) 300,000 years ago we didn't have institutional science and, well, this lovely forum.

With all that in mind, I do not believe that debates such as we have here are pointless. I suppose I'm just largely in the overcoming-bias camp: we know a certain amount about the kinds of mistakes people are inclined to make without noticing, ourselves included, and we can improve our performance by being on the lookout for those mistakes. Logic was a big damn step, but there are others.

Shall we talk about pandemic ethics, now? I believe I understand your overall approach quite a bit better than I did a few days ago, so I'm curious to see if I can actually apply any of this to the questions at hand.

in Disney's Sleeping Beauty, the weapon with which Prince Phillip slays Maleficent (in dragon form) is the Sword of -- wait for iiiiit -- Truth. — Srap Tasmaner

Ha! Indeed. Laying on the clichés quite thick in that one.

Actually, yes, that was what I was thinking. Quite short-term gains in efficiency, or gains within a department, could be overall inefficient, or in the long-term inefficient. It's a danger hierarchies are prone to by nature. Examples from the business world are endless. — Srap Tasmaner

Are they though? I don't mean the short-termism in general (that's lamentably common!), but Truth? Did that notion help them in the long-term? Or was it just a more efficient model? If a model produced better long-term gains but was no more 'true' (or even perhaps less true) would we have any reason left to prefer truth? Aesthetics, perhaps? Or do you think, perhaps, it's impossible for a model to produce good long-term gains without being closer to the truth?

evolution selects for an organism to have certain capacities that meet a need, but that doesn't mean those capacities are limited to meeting that need. We didn't evolve to be able to play baseball, but we do. I even have a pet theory that language is an accident, that we got an upgrade on our signaling ability that is far greater than any species could ever need. — Srap Tasmaner

Definitely with you so far.

we can make the attempt. — Srap Tasmaner

You hinted at the problem, and we've talked about it before, I think, but the facts we have simply over-determine most theories that intelligent people can come up with. The best we can hope for is to rule out the loonies. Something like this coronavirus situation, despite the way my numerous detractors paint it, there's just no way of pinning down any truth of the matter. Most (sensible) theories can be supported by the range of facts available, so all discussion can show us (if we assume it's anything more than storytelling - of which I've yet to be fully convinced) is the manner in which people muster their particular facts to support their particular theory...unless I'm one of the loonies that need ruling out...!??

Shall we talk about pandemic ethics, now? I believe I understand your overall approach quite a bit better than I did a few days ago, so I'm curious to see if I can actually apply any of this to the questions at hand. — Srap Tasmaner

Yeah, let's go for it.

but Truth? — Isaac

Sorry, I must not have been clear enough: I was only talking about efficiency there. (I was deliberately passing over the other stuff you talk about there, the asymptote of truth and all that.)

over-determine — Isaac

"underdetermine" I believe you'll find.

Yes, so let's grant the Quine-Duhem thing. There are still grounds for distinguishing different kinds of theories, or distinguishing one framework from another, even if "adequacy to the known facts" is no help.

I'll give a bone-headed example. I require a rule to explain this sequence

1 2 3 4

Here are two of the many rules that are adequate to the facts we have so far:

(A) The first number is 1; the nth number is the successor of the (n-1)th number.

(B) The first number is 1, the second is 2, the third is 3, and the fourth is 4.

For some purposes these rules are both good enough, but for some (A) is better and for others (B). For a lot of the stuff we care about around here, (A) is the hands-down winner.

I believe there is a lot we can say here, and I've made no special study of theory-building. There are natural virtues to look for though: robustness, generality, extensibility, "explanatory power" etc. And that's before we consider the consonance of this theory with other theories competing in their domains, the construction of theoretical frameworks, of research programs, and so on.

In short, I can take Quine-Duhem as given without throwing up my hands. In fact, it looks like an advantage that we can try new frameworks without giving up whatever progress we've made. In the longest possible run, there will still be an infinite number of theories available to explain everything that's ever happened, but so what? By then, we should be satisficed (see what I did there?) with any one of those. At that point, all bets are off anyway. Until then, so long as we are still in the process of figuring things out and there's still new data coming in, there are more things of interest than adequacy to the currently known facts.

With that out of the way, I'll start writing about coronavirus, but it should already be clear that I'm not going in assuming it's all just stories and all stories are equally good -- or whatever the right caricature of your position is. I do worry though that I may not know enough to make the distinctions others could, but I will try.

Something like this coronavirus situation, despite the way my numerous detractors paint it, there's just no way of pinning down any truth of the matter. Most (sensible) theories can be supported by the range of facts available, so all discussion can show us (if we assume it's anything more than storytelling - of which I've yet to be fully convinced) is the manner in which people muster their particular facts to support their particular theory — Isaac

For simplicity, starting with a single issue: whether to get vaccinated.

I don't know if the following is any good at all -- it's all off-the-cuff -- and it's not perfectly obvious how it connects to our recent more abstract exchange, but it has the virtue of going directly at the main question...



What's interesting, and with any luck helpful, here is that this is not the typical case of ethical judgment. In our case, everyone forming such a judgment has faced the same choice themselves.

That means there are two obvious options, which may or may not be important:

1. Approve of making the same decision I did; disapprove otherwise.
2. Approve of following the same process I did; disapprove otherwise.

For people who want both 1 and 2, there's a potential quandary if someone uses the same process but with a differing result. Presumably that indicates they used differing inputs. They shouldn't do that, hence

3. Approve of using the same inputs I did; disapprove otherwise.

If I did the math right, 2 + 3 = 1, unless the procedure in 2 is stochastic.

This might not seem like much of a basis for an ethical judgment, but if you presume everyone facing this choice does so with the intention of behaving ethically, of judging their own decision to be an ethical one, it's not all that crazy.

Can we make the just-like-me approach fail? Is that even possible, if I've set up my criteria this way?

I'm going to cheat now, because it looks to me like the weak spot is 3 (which in turn will tend to weaken 1). This is the weak spot because "inputs" looks way too big: that's not just what you read in the news, or what you read in scientific journals, if you do that sort of thing, or what you may have experienced either personally or professionally; it's also you, your personal health and your circumstances. If you're allergic to something in the vaccine, you can't take it, even though I can, and there's no way I can ignore that and be ethical.

So how do we account for that sort of difference with a rule as simplistic as the rules above? Remember, we're only doing this -- only making these ridiculous rules -- because in this case everyone judging another's decision has had to make exactly such a decision themselves, and that's not the usual case. We're not crafting the General Rules of Ethical Behavior; we're letting people leverage the work they already put in making their own decision to reduce the burden of judging others. Because we can.

So far as I know, I am not allergic to anything in the vaccine. Does someone who is have to make the same choice I did about whether to get vaccinated? That looks like a definite "no" to me. They had no choice. What does that mean for our rules? Have we succeeded in forcing failure? Rule 3, being overbroad, fails, and thus many instances of 1? (Some people might just plump for 1 straight-up, and they're fine.)

I don't think so. I think you get to keep just-like-me and simply exclude the allergic. They didn't face the choice I did, made no decision like or unlike mine, and I judge them not.

How far can we go with this faced-the-same-choice-I-did business? Do we expect the circle to shrink and shrink and shrink until it's only me that faced the same choice I did? I don't see why. But I admit it is now unclear whether the hard part -- which we have made shockingly easy for ourselves so far -- is reaching an ethical judgment, or deciding who is subject to our judgment.

For a concrete example, suppose I am obese and have diabetes. I am at risk of getting seriously ill and needing hospitalization if I get infected; for simplicity, let's say I consider it an ethical duty to minimize the risk of serious illness** so I get the vaccine. Now let's suppose someone else, call him "Isaac", has neither of the risk factors I do and is generally in very good health; Isaac chose not to get vaccinated. Do I count Isaac as facing the same choice I did? He had to decide whether or not to get vaccinated; he may have exactly the same goals I do of not getting seriously ill and needing to be hospitalized; he may have weighed the odds just as I did using the same cutoff for acceptable risk I did (this would be a rule 2 sort of thing) -- but wait a minute! What odds was he weighing? Were they the same ones I was weighing?

You get your choice here. I'm inclined to say yes, because it captures the point that we get whole columns of odds from our local public health officer, broken down by risk factor, maybe age, and so on. I kinda want those to count as one thing because they have one source and we acquire them as one thing. More tellingly, the odds are not exactly a fact about you; that certain odds apply to you, and certain odds don't, is a fact about you.

Which brings me right to the next bit: Isaac weighed the same odds I did; he selected from those columns of odds the ones that apply to him, just as I did; but the particular odds he selected were different because he's different. There is an exact point where -- even though he followed the same process I did with the same external inputs -- because the process involves direct reference to the decision maker, he diverged!

What do I do about the Isaac case? Remember, I don't really want to say that he failed to use the same inputs as I did (that I used me, and he used Isaac) and so is subject to my judgment but fails rule 3: he read exactly the same odds sheet I did, and I want to call that responsible and ethical. But when he did, and checked for his risk factors, he found different odds applied to him.

That's a problem because I approve of Isaac's inputs, and I approve of his process, so I should approve of his choice, but his choice was different from mine, so how can I approve? The whole point of 2 + 3 = 1 is that it's how I judge my own decision to have been ethical. If I have to let Isaac slide, I have to give up something: either I have no basis for concluding that my own decision was good (before it was because I did 2 and 3 right), or I just give up all the rules past 1 and disapprove of Isaac.

I can plump for same-decision-as-me, but suppose I really like the 2 + 3 = 1 approach; can I rule that Isaac, because his odds were different, did not face the same choice I did and is not subject to my judgment, just as if he were allergic to the vaccine? I think that's a cop-out. You save the model from failure only by pushing the failing case outside the domain of application.

Besides, maybe we don't want to give up judging people with different odds; maybe Isaac is going through the same thought process we are and wants to be able to tell people with multiple risk factors, people like me, that the right decision for them is to get vaccinated.

Where we stand: we have forced the complete version of just-like-me, with all 3 rules, to fail. I have to approve of Isaac's decision because of rules 2 and 3 -- he did the same thing I did; but I have to disapprove of Isaac's decision because of rule 1 -- he didn't do the same thing I did.

Our options:

• add more rules
• give up everything past rule 1 -- looks bad, that's like just defining your decisions to be ethical
• give up rule 1 but keep 2 and 3 -- appealing because I still count as ethical, but a little weird that my actual decision drops out -- wasn't the whole point to judge the decision itself, mine and Isaac's?
• give up just-like-me altogether -- too much work, or at least too soon

** You could read this as simple rationalist egoism, but there are alternatives: maybe I consider life a gift deserving respect and conservation, and that includes my own, or maybe I feel I have a duty to those who need or care about me, maybe I'm concerned about being burden on the healthcare system. Positing this as my goal is simple and we can treat the moral and rational approaches the same.

First things first...

I was only talking about efficiency there. (I was deliberately passing over the other stuff you talk about there, the asymptote of truth and all that.) — Srap Tasmaner

Understood.

"underdetermine" I believe you'll find. — Srap Tasmaner

Ah, yes. Facts underdetermine, theories are overdedetermined. I'm wondering now how many times I've written them the wrong way round before!

There are natural virtues to look for though: robustness, generality, extensibility, "explanatory power" etc. And that's before we consider the consonance of this theory with other theories competing in their domains, the construction of theoretical frameworks, of research programs, and so on. — Srap Tasmaner

The problem (as I see it) with all of those is this; that 'theory A is robust' is itself a theory. It'd have facts (the various properties of theory A) and those facts would be mustered in various ways to support the theory that 'theory A is robust'. So why wouldn't those facts also under determine the theory?

Same can be done with coherence, generality, extensibility, "explanatory power" etc. They are all theories about theories and so subject to the same problems.

so long as we are still in the process of figuring things out and there's still new data coming in, there are more things of interest than adequacy to the currently known facts. — Srap Tasmaner

'Of interest', however, I can definitely get behind.

So...

Where we stand: we have forced the complete version of just-like-me, with all 3 rules, to fail. I have to approve of Isaac's decision because of rules 2 and 3 -- he did the same thing I did; but I have to disapprove of Isaac's decision because of rule 1 -- he didn't do the same thing I did. — Srap Tasmaner

I like the way you've laid it out. It's not a way I'd have looked at things at all, which itself is interesting. For me, (1) doesn't even figure, I assume there are generals, soldiers, nurses, and engineers in any army and so the idea that someone's role could be judged by how similar to mine it is doesn't seem viable off the bat. But then that's my story-telling taking over (it takes an ensemble cast to fill out a proper play) and we agreed to try and avoid that, so I'll frame it another way...

What if we use Wittgenstein's "stand roughly here"? What does it mean to have made 'the same' decision as me? Surely not to have had the same nurse inject the same vaccine in the same hospital? We have to set some 'rough' boundaries around what constitutes 'the same', yes? I think here is were I find it difficult to escape the psychologising - the choice of easy to identify badges for what constitutes 'the same' is a group dynamic based choice. They're easier to use to identify in group and out group. We could have chosen something vague (but more useful) like 'health robustness' as what constitutes 'the same' - "Did Isaac's decision making result in 'the same' level of general health robustness'.

I think the problem is that you're missing a (0.5) which is the jointly held objective. It should go

0.5 - Approve of having the same objective I do, disapprove otherwise
1. Approve of making the same decision I did; disapprove otherwise.
2. Approve of following the same process I did; disapprove otherwise.
3. Approve of using the same inputs I did; disapprove otherwise.

But then we can lose (1) altogether

0.5 - Approve of having the same objective I do, disapprove otherwise
2. Approve of following the same process I did; disapprove otherwise.
3. Approve of using the same inputs I did; disapprove otherwise.

Which works, I think, except it suffers exactly the same problem in abandoning (1) in your original scheme...

give up rule 1 but keep 2 and 3 -- appealing because I still count as ethical, but a little weird that my actual decision drops out -- wasn't the whole point to judge the decision itself, mine and Isaac's? — Srap Tasmaner

This, really is our conundrum. Intent matters in ethics. But we can't judge intent, we can't see it, so we have to use a proxy. Virtues are usually a good proxy, but they're the result of long tradition (a long tradition of story-telling - sorry, couldn't resist). We don't have such a long history here. That's why, way back, I brought up the Wakefield/MMR scandal, I think it's our nearest story on which to judge intent (by the proxy of actions) - the sensible, community-minded took the MMR (good), those more interested in appearing 'niche' didn't (bad). I know I keep circling back to stories and I promised I wouldn't. Let me try harder...

To judge intent is essential in an ethical judgement because the actual decision might be very circumstantially specific, but we can't see intent. We have two options;

1. Judge good intent by other means - trust, general character, other decisions...(très system 1)
2. Analyse the circumstances carefully - we do the work to judge intent by getting into those specific circumstances and seeing if they do or don't lead to the outcome - if there's no way we can get the moral outcome from the circumstances, then the decision cannot have been made with good intent...(très system 2)
3. Judge good intent by interrogation - is there a plausible, rational argument that can be presented linking the specific circumstances (your 2 and 3) to the moral objective. If there is (and we're charitable) there's no reason to doubt good intent.

Passing by the theory of theorizing, with some effort.

It's not a way I'd have looked at things at all — Isaac

Nor I!

Not all that happy with the result, but glad I did it. I was in a mode of trying to think through it in the most bone-headed way I could imagine, and it struck me that the absolute simplest way to judge someone else is by whether they do the same thing I do. (It's not impossible that has actually happened in this thread.)

As you say, we get pretty quickly into what's to count as "the same", but I'm still glad I started with (1) because it does something helpful, or, rather, it would have if I had been a little more careful: it leaves no room for akrasia. I think this is why I was reluctant to jettison it at the end of the post -- you don't get credit for having the right intentions but for acting with the right intentions. I really want to keep that, despite my oh-so-modern inclination to redefine the intent of the one who doesn't act. (He must have some other preference that's even stronger, blah blah blah.) What if we don't rush to treat intentions as a sort of theoretical generalization of actions, but honor the traditional recognition that between the intention and the act there falls a shadow?

my story-telling ... we agreed to try and avoid that — Isaac

Oh no, I just said I wasn't going to do it for you. You do you. I have Lakoff (who's a challenge for me, temperamentally) and Goffman in my to-read-soon-ish pile. I'm interested in your narratives, it's just not one of my tools -- oh and I'm slightly allergic to the word "narrative" but I'll get over it.

Back to business.

The big virtue of same-as-me as a strategy is that it's dirt cheap. It even has a sheen of reasonableness in this case that it usually lacks -- usually it looks like bigotry, to be honest -- because we are all of us facing some version of the same choice.

It's such a dead simple strategy that I wonder if it isn't always step 1, but one we've learned to pass over so quickly we almost miss it. So what I'm interested in is how and when we say "Just like me unless ..." or "Just like me except ..."

Here's a sort of sitcom example. Older conservative businessman and a younger female colleague heading to a meeting; older guy wearing standard conservative suit, young woman dressed like she's going out for drinks with friends (whatever that looks like). What the older guy wants to say is, essentially, "Why aren't you dressed like me?" even though that's ludicrous on its face (hence sitcom -- Brooks Brothers didn't have that suit in my size). So what he has to say instead is something like, "Why aren't you dressed like me, but in a way that's appropriate for you?" Or, "Why aren't you dressed the way I would dress if I were a young woman?" Yet another way to put this might be: "You can't dress exactly like me, because you're a young woman, but why didn't you dress as much like me as you could?" And hovering in the background is the fully generalized version: you're not me, but why aren't you as much like me as you can be? (And possibly there's a weird double-judgment under that: why don't you want to be as much like me as you can be? What about me do you disapprove of?)

Sadly, heading to work -- more later.

Quick note on the methodology here.

The idea of the bone-headed approach is to resist using the abstractions (intent, values) that have already accumulated here, so that we can catch abstraction in the act. We want to see what motivated the step of abstraction, or what forced it. What is the precise function of the step of abstracting, what purpose or need does it serve?

Around here that might be, what failure would force me to consider an abstract element of my action, and of yours, called the "intention"? That's more work, so why do we do it? (That's actually an open question at the moment. I could see an argument that checking whether we're the same in one way is cheaper than checking if we're alike in general, since that's checking a huge number of component identities. But that assumes a lot of abstracting already done. It just sounds too textbook-ish to be true.)

(( It used to be common to say that ancient cultures did not consider your intention at all when judging your actions right or wrong. Sleeping with your mother is wrong even if you don't know it's her. And, the story goes, it was Christianity that ushered in the era of obsessing over intentions. Don't know if there's any truth to any of that. If so, we're trying to catch ontogeny recapitulating phylogeny. ))

One further note on my example: it should be clear enough how full-on bigotry works here; if certain steps of abstracting are not available, you might quickly conclude, "You're a woman (or not white, or the wrong religion, etc.) so there's nothing you can do to be enough like me," for some value of "enough".

But we'll be getting to putting values on "enough" eventually, I'm sure.

the most bone-headed way I could imagine, and it struck me that the absolute simplest way to judge someone else is by whether they do the same thing I do. (It's not impossible that has actually happened in this thread.) — Srap Tasmaner

More than once I think.

you don't get credit for having the right intentions but for acting with the right intentions. — Srap Tasmaner

True. My 'Three Point Plan' is no improvement because it only contains intentions, and actions are the best proof we have of intentions. But this is why I provided some tentative means of judging intent, albeit roughly.

I have Lakoff (who's a challenge for me, temperamentally) and Goffman in my to-read-soon-ish pile. — Srap Tasmaner

Excellent. Goffman on Frame Analysis and Lakoff on metaphor, I hope. Their best contributions.

I'm slightly allergic to the word "narrative" but I'll get over it. — Srap Tasmaner

Yes, it's got a slightly 'social science' taint to it now...sorry. I have to still use it though, I'm supposed to be a professional, I can't go around saying 'story', I sound like a five year old.

possibly there's a weird double-judgment under that: why don't you want to be as much like me as you can be? What about me do you disapprove of?) — Srap Tasmaner

Yes! This gets very much to the heart of what these judgements are doing. But it's selective - back to stories again (see how I avoided 'narrative'). It's those who ought to be playing the same part as us but who appear to be ad-libbing their lines, or reading from a different script. It's nit that we actually want everyone to be like us (would you?!), but we want the way we play out our roles to be predictable and secure, we don't want similar roles played too differently. Your sitcom example wouldnot have worked if she was an escort.