Simplicity (2016)
plato.stanford.edu
plato.stanford.edu
One argument that I've got from biology is: The simplest possible system (code, living organism) is fragile. Any change, however minor, is likely to break it. A nice chubby complex system, on the other hand, is likely to be more resilient. You may change it a bit and it's still likely to work somehow.
You should take a look at tardigrades then (aka water bears), which are approx. 1mm in length, obviously less complex than most living creatures, yet are some of the most resilient animals on earth; highly resistant to extreme temperatures, pressure, radiation, etc.
One way to look at resilience: a relationship between the system and its environment. So when you say resilient, you need to specify, resilient to what? I think a resilient system, is neither too complex, nor too simple, but rather, adapted to its environment.
On a related(?) path see https://en.wikipedia.org/wiki/Exaptation
1. A lot of people have an idea that “Analytic Philosophy” is a well defined school with some kind of shared theses and well-defined approach. That hasn’t been true for a long time, if it ever was. I’m not saying that there’s zero distinction between so-called analytic philosophy and any number of continental schools of thought, but if you think either side is homogenous, you’re wrong.
2. The SEP includes articles on non-analytic figures and movements by sympathetic authors. Those figures include historical philosophers, continental philosophers and non-Western philosophers.
Put another way: to "lift" philosophy into the domain of science would involve defining/formulating the guiding principles of sound/"good" science in their own terms.
(I was educated in the analytic tradition.)
[1]: It's worth noting that this only covers the philosophies of science and mathematics, not things like ethics, aesthetics, existence, &c. I think we can defend investigating the latter subjects as a consequence of science's inability to answer normative questions -- science can only tell us what is, not what ought to be.
It's worth noting that there are some who disagree and have some good arguments in support of that[0], and that philosophy as Wissenschaft was prevalent until relatively recently. To me, the continental/analytic distinction is a distinction in the methods of inquiry, if it is a valid distinction at all.
Per your link: I wonder what the author of that page thinks of the Gettier problems and whether they constitute "false knowledge."
Honestly I always felt thiswas a smokescreen. Science can totally explore philosophical and moral questions once axioms have been laid out. Science is illegitimate to lay down these axioms, but so are philosophy, spirituality, religion, or any other justifications.
Studying what is does not forbid to study morality, its ramifications, justifications and consequences as moral systems do exist, as human constructs. Can you deduce humanism from purely utilitarian axioms? Does it lead to similar systems? How about when one adds bounded rationality? These are scientifically answerable questions as well as questions that I would consider philosophical.
The fact that modern academic philosophy feels unbound to scientific rigor in fields when it does not have to be so is totally unjustified IMO. And I think that one can obtain much more interesting debates if we keep the tested epistemological tools of science to infer complex normative consequences from common axioms.
Where do you we propose that we get our moral axioms from, then? I don't mean that as a dismissal, just to observe that rejecting "any other justifications" seems to leave us incapable of producing real/universal moral axioms. That spells trouble for domains like science, which are generally in the business of distinguishing the real things in the world from the fake ones.
> Can you deduce humanism from purely utilitarian axioms? Does it lead to similar systems? How about when one adds bounded rationality? These are scientifically answerable questions as well as questions that I would consider philosophical.
I agree that they're philosophical, but I don't think they're scientific as posed: "deduce" suggests a sort of logical proof, one that we'd probably accept a priori (i.e., without needing to experiment or develop any sort of knowledge of the world). Similarly, "similar" in what sense? I don't think science has (so far) given us good answers to the problems of universals/categories/platonic forms, and I think that doing any sort of moral science would require that we first tackle that.
> The fact that modern academic philosophy feels unbound to scientific rigor in fields when it does not have to be so is totally unjustified IMO.
I think it's worth noting that scientific rigor is not the only sort of rigor out there -- philosophers, especially analyticists, pride themselves on developing clear definitions and using formal logical systems for argumentation. The vast majority of philosophers would not accept gut feeling as a substitute for sound argumentation, nor would they shy away from scientific results as inductive evidence for a claim. I think that is the sort of rigor we should demand from philosophy, not that of science.
In the same way that science can tell you how to create chocolate or vanilla aromas but can't tell you which one tastes better.
> "deduce" suggests a sort of logical proof, one that we'd probably accept a priori
Indeed, there is a lot of work to do to translate the tools of logic into the domain of morality, define the conditions and limitations of the methodology, and possibly at some point do experiments or observation to check if the model matches reality.
Thing is, we can totally approach these questions scientifically and the fact that philosophers do not focus on making the tools to unleash the full power of science over these questions is hard to justify.
I don't agree with this, but I'll go with it for the sake of argument. So: if we think that moral axioms are subjective, where does science enter into it? How could we even begin to do science, the field most invested in picking out real things in the world, on a set of axioms that we concede don't map to any objective truth?
> Indeed, there is a lot of work to do to translate the tools of logic into the domain of morality, define the conditions and limitations of the methodology, and possibly at some point do experiments or observation to check if the model matches reality.
Reasonable. But what epistemic principle tells us that affirming or disaffirming observations are important at all? How can we derive that principle, if it indeed exists? If we're modeling some system of morally relevant actions, what metric are we using, and where did we get it from? You mention "reality," but we haven't overcome the tension between subjective moral axioms and an objective, scientific, decision procedure for moral choice.
In the same way that science enters into the making of pink or blue paint. It does not say which color is the prettiest, just how to manufacture them and the characteristics they have.
Similarly, science can tell how morality systems appear, are transmitted, what their consequences and eventual contradictions are: You think we should have equality of chances? Here are the likely outcomes given our current society. You think we should have equality of outcomes? Here are the factors to adjust for.
Right now, deep learning programmers are having very interesting philosophical discussions on ethics because their models, trained on real datasets, end up acquiring the bias of the real world. E.g. a CEO is more likely to be a man and a nurse more likely to be a woman.
They are struggling with how and when correct the biases and when leaving them. We are now rushing laying the theoretical groundworks for a practical theory of ethics because we are the one who will be blame for what our models learn and because, quite frankly, philosophers have not done their jobs there.
> But what epistemic principle tells us that affirming or disaffirming observations are important at all? How can we derive that principle, if it indeed exists?
That's another conversation altogether, and in our case irrelevant: we are doing science here, even on a philosophical subject. The whole point of science is to build a predictive model of the world that is coherent with observations. If you are doing science, you care about reality matching your model.
I know philosophers like to ponder this question a lot instead of thinking about harder practical problems but really, this is a distraction. I find it very reasonable to assume that we care about reality and to ignore solipsism for now.
> If we're modeling some system of morally relevant actions, what metric are we using, and where did we get it from?
The question of metrics is indeed very important, and economics and sociology have that discussion. They use imperfect metrics and build a model using them, while in the same time trying to get better ones. Specifying a moral system in terms of objective metrics is indeed a first step. Why has there been so little work on this?
Things like life expectancy, likelihood of violent death, reported happiness or freedom, HDI, are all metrics that can be used to evaluate if a model is succeeding at its goals.
> You mention "reality," but we haven't overcome the tension between subjective moral axioms and an objective, scientific, decision procedure for moral choice.
I think this one is simple, (assuming I understand correctly what disturbs you). Do as in math: state what is true, given some explicitly given assumptions. "If we consider only positive integers, then a^3+b^3=c^3 has no solution". "If we consider only numbers in Z2 then 1+1=0"
Unless we're committed to the idea that moral claims have the same sort of weight as preference claims about colors, I don't think this analogy holds water. Science can't tell us how to make value-independent morality in the way that in can tell us to make value-independent paint, because moral systems are themselves fundamentally normative.
> Similarly, science can tell how morality systems appear, are transmitted, what their consequences and eventual contradictions are: You think we should have equality of chances? Here are the likely outcomes given our current society. You think we should have equality of outcomes? Here are the factors to adjust for.
It can probably tell us how norms appear, are transmitted, &c. You're burying the more serious question here: why am I posing this thing (equality of opportunity) in this way ("good") to begin with?
> We are now rushing laying the theoretical groundworks for a practical theory of ethics because we are the one who will be blame for what our models learn and because, quite frankly, philosophers have not done their jobs there.
I take issue with this: practical ethics is one of the shining achievements of 20th and 21st century philosophy (along with advances in the phil. of math and science that have enabled this conversation). IRBs, harm reduction policies, the UN Declaration of Human Rights, Effective Altruism, and ethical consumerism are all direct products of practice-oriented moral philosophy. The fact of the matter is that computer scientists and engineers, including ones in my own subfield, simply haven't done their homework.
> The whole point of science is to build a predictive model of the world that is coherent with observations. If you are doing science, you care about reality matching your model.
According to who? That's not Quine's science, or Popper's, or Carnap's. It's certainly not Newton's or Aristotle's. It's not immediately obvious to me what "science" even is beyond the vaguest senses (it's certainly epistemic, and just about everybody agrees that it involves knowledge of real things).
And, to be clear, this isn't solipsism: I am not doubting the existence of science, or its ability to produce truth. I'm just pointing out that we haven't established our scope yet.
> Specifying a moral system in terms of objective metrics is indeed a first step. Why has there been so little work on this?
There's actually been a great deal of work on this: Bentham and Mill both sought to structure their utilitarian systems on formal (and quantifiable!) evaluations of pleasure and pain. Singer follows suit, albeit with fuzzier orderings rather than quantities. Kant's metaethics are essentially a souped-up law of non-contradiction. Reasoning about natural rights follows a similar pattern.
> Do as in math: state what is true, given some explicitly given assumptions. "If we consider only positive integers, then a^3+b^3=c^3 has no solution". "If we consider only numbers in Z2 then 1+1=0"
This would involve a predication of any metaethical system, e.g. "if we accept Mill's utilitarian calculus, then the right thing to do in situation X is Y." But here's the problem: we don't all accept Mill's utilitarian calculus, and those of us who don't aren't likely to be persuaded in the same way that considering only the natural numbers in a mathematical problem might. When it comes down to actually acting, we're back where we started: why ought I accept Mill's utilitarian calculus?
One of the problems of modern academic philosophy, though, is that it does think it is more legitimate than others at making normative statements. Its more relaxed ways of justifying statements make it easier to disguise opinions as genuine philosophy, however I do believe that it is something else totally.
> According to who?
To anyone doing science actually. If you try to publish something that improves a predictive model of the world, you'll get published. If you just try to publish unjustified speculations, it is far less likely.
> I'm just pointing out that we haven't established our scope yet.
That's the problem I have in almost all discussion I have done with academic philosophers: they want to hide behind semantics problem to avoid tackling actual problems. If you are uncomfortable calling it science because you have a different definition of the word, fine, call it what you want. After all, I have my own definition of philosophy too. Thing is, it does not stop me from wanting to have a practical model for ethics.
> There's actually been a great deal of work on this
I have seen several people claim this but none to link to convincing freely available material. Any field of science will have understandable introductory material online, if you have one on these formal ethics systems, I'd be curious to read it.
> The fact of the matter is that computer scientists and engineers, including ones in my own subfield, simply haven't done their homework.
Since when is ethics within the scope of CS and engineering more than philosophy? When developers need notions of maths, they turn to mathematicians. When they need physics, they turn to physicists. When they needed ethics, they had to come up with answers themselves. So you are right, it is becoming an engineering discipline, but that's a shame for academic philosophy.
> When it comes down to actually acting, we're back where we started: why ought I accept Mill's utilitarian calculus?
Really, the same way we do when we choose between blue and pink paint: based on our personal preferences, which come from our culture and past experience.
Convincing someone to use moral predicates is not a job for science nor, IMO, for philosophy. This is a battleground that belongs to politics, religion, arts, culture. Science and philosophy can provide good arguments to them, but can't justify a specific axiom without using other axioms.
Yes, because this would entail moral relativism: I would have to accept that for any course of action that I think is right, there might be any number of equally right but diametrically opposed actions. This undermines the entire point of a moral system, which is to be universal.
> Its more relaxed ways of justifying statements make it easier to disguise opinions as genuine philosophy, however I do believe that it is something else totally.
I apologize if this is incorrect, but I think this is coming from a position of ignorance. You cannot just wave your hands in (analytic) academic philosophy and expect people to accept, must less respect, your positions. You need to present a proof, containing formal methods, that your claim follows from your premises and that your premises are true. I'd wager that the majority of moral philosophers accept the validity of moral systems they don't think are correct, but not the soundness.
> If you try to publish something that improves a predictive model of the world, you'll get published. If you just try to publish unjustified speculations, it is far less likely.
But wait: this isn't a definition of science, it's just an efficient means for getting it published. It also doesn't cover science that produces new models. If every major science journal required me to sign a letter affirming the existence of God, would be we comfortable with that being a defining characteristic of the field?
> If you are uncomfortable calling it science because you have a different definition of the word, fine, call it what you want
I don't have a different definition, at least not necessarily. What I'm pointing out is that you're placing a lot of value on this thing we call science, and yet we haven't really established what it is. Is it synonymous with truth? How do we go about doing science on purely formal systems, like mathematics? What do we do with scientific results that are later invalidated?
Essentially, I don't think science is the only sound epistemic process. In the trivial case, logic is another one (and one potentially more important). Less trivially, I would claim that morality is a third.
> Any field of science will have understandable introductory material online, if you have one on these formal ethics systems, I'd be curious to read it.
J. S. Mill's Utilitarianism is an easy read, and is freely available online. Peter Singer also makes most of his work available. Kant's Groundwork for the Metaphysics of Morals is not an easy read (especially not with free older translations), but he is also available online. Similar for Aristotle's Nicomachean Ethics, which is probably somewhere between the other two. Together, these three make up the dominant meta- and ethical positions in the field today.
> When they needed ethics, they had to come up with answers themselves.
My point above was that they didn't have to. More concretely, they haven't, in at least one case: I taught an undergraduate course on ethics in cybersecurity, and used ethics material directly from philosophy.
When I say "haven't done their homework," I'm not referring to a need to build their own ethics -- I'm referring to the prevailing arrogance that allows them to dismiss moral philosophy without even understanding it first.
> This is a battleground that belongs to politics, religion, arts, culture.
But these are all things that philosophy either built or tempered!
Our political systems (in the West) are the direct consequence of 17th, 18th, and 19th century political philosophy, all of which have their own lineages. Most religions make substantial philosophical claims, and even respond to the philosophical challenges presented both inside and outside their orders. The field of aesthetics makes claims like "art is in the eye of the beholder" sensible to begin with.
Now, there's an obvious sense in which what you say is true: it does ultimately come down to my personal preferences. But again, this doesn't tell me what my personal preferences should be. We could appeal to the worlds of politics, religion, &c for answers, but these all ultimately make normative claims based on their philosophical underpinnings. Why not cut out the middleman?