In the comments on Soulless Morality, a few people mentioned contributing to humanity's knowledge as an ultimate value.  I used to place a high value on this myself.

Now, though, I doubt whether making scientific advances would give me satisfaction on my deathbed.  All you can do in science is discover something before someone else discovers it.  (It's a lot like the race to the north pole, which struck me as stupid when I was a child; yet I never transferred that judgement to scientific races.)  The short-term effects of your discovering something sooner might be good, and might not.  The long-term effects are likely to be to bring about apocalypse a little sooner.

Art is different.  There's not much downside to art.  There are some exceptions - romance novels perpetuate destructive views of love; 20th-century developments in orchestral music killed orchestral music; and Ender's Game has warped the psyches of many intelligent people.  But artists seldom worry that their art might destroy the world.  And if you write a great song, you've really contributed, because no one else would have written that song.

EDIT: What is above is instrumental talk.  I find that, as I get older, science fails to satisfy me as much.  I don't assign it the high intrinsic value I used to.  But it's hard for me to tell whether this is really an intrinsic valuation, or the result of diminishing faith in its instrumental value.

I think that people who value rationality tend to place an unusually high value on knowledge.  Rationality requires knowledge; but that gives knowledge only instrumental value.  It doesn't (can't, by definition) justify giving knowledge intrinsic value.

What do the rest of you think?  Is there a strong correlation between rationalism, giving knowledge high intrinsic value, and giving art low intrinsic value?  If so, why?  And which would you rather be - a great scientist, or a great artist of some type?  (Pretend that great scientists and great artists are equally well-paid and sexually attractive.)

(I originally wrote this as over-valuing knowledge and under-valuing art, but Roko pointed out that that's incoherent.)

Under a theory that intrinsic and instrumental values are separate things, there's no reason why giving science a high instrumental value should correlate with giving it a high intrinsic value, or vice-versa.  Yet the people here seem to be doing one of those things.

My theory is that we can't keep intrinsic and instrumental values separate from each other.  We attach positive valences to both, and then operate on the positive valences.  Or, we can't distinguish our intrinsic values from our instrumental values by introspection.  (You may have noticed that I started using examples that refer to both intrinsic and instrumental values.  I don't think I can separate them, except retrospectively; and with about as much accuracy as a courtroom witness asked to testify about an event that took place 20 years ago.)

It's tempting to mention friends and family in here too, as another competing fundamental value.  But that would demand solving the relationship between personal values that you yourself take, and the valuations you would want a society or a singleton AI to make.  That's too much to take on here.  I want to talk just about intrinsic value given to science vs. art.

Oh, and saying science is an art is a dodge.  You then have to say whether you value the knowledge, or the artistic endeavor.  Also, ignore the possibility that your scientific work can make a safe Singularity.  That would be science as instrumental value.  I'm asking about science vs. art as intrinsic values.

EDIT:  An obvious explanation:  I was assuming that people here want to be rational as an instrumental value, and that we should find the distribution of intrinsic values to be the same as in the general populace.  But of course some people are drawn here because rationality is an intrinsic value to them, and this heavily biases the distribution of intrinsic values found here.

New Comment
36 comments, sorted by Click to highlight new comments since:

All you can do in science is discover something before anyone else discovers it

I'm not sure there isn't an equivalent of this in the art world. Ok, no-one else would have created exactly the same piece of art as you, but is that what's really imporant? All you can do in, say, literature, is write a book that people read instead of some other book. Given that no-one is ever going to read every book, see every play or listen to every song, the marginal value of your artwork to the world is not how much people enjoy it (or how valuable it is in whatever sense you think art is valuable), but how much people enjoy it (how valuable it is) compared to their second best work already available.

Yes, and I almost addressed that in the original post, but decided the explanation would distract. My impression is that there's plenty of room for great works; and people still read and listen to crap. The kicker is that if what you care about is people who read your book, then you probably value your book for its instrumental value. Yet all authors care about whether people read their books. Does this mean no authors place intrinsic value on their works? I think the answer is more complicated than that, and although its answer is probably at some point necessary to get further towards an answer to the question I'm asking in this post, it is much further down the road than what I'm asking here

"If what you care about is people who read your book, then you probably value your book for its instrumental value"

How much would you value the following situations?

  1. By odd coincidence, a brilliant work of literature forms spontaneously in the Andromeda galaxy. No one ever reads it.

  2. An author writes a brilliant work of literature, then immediately burns it and shoots himself, so that it is lost to all time.

  3. An author writes a brilliant work of literature, then buries it and never tells another living soul.

  4. An author writes a brilliant work of literature. He lets one other person read it, then buries it and never tells another living soul.

  5. A Chinese author writes a brilliant work of literature. Millions of Chinese people read and enjoy it, but is never translated into English and neither you nor anyone you know ever hear about it.

  6. An English author writes a brilliant work of literature. You and many others read and enjoy it.

Through situations like these, I've decided that what I value in art has something to do with people enjoying it. I place minimal value on 1-4, but high value on 5 and 6. I don't think I believe in the simple form of 'only this one thing called happiness is important' that is vulnerable to the wirehead trick, but I do think that everything valuable is linked to a form of what we would call happiness. I would say my terminal value is "enjoyment of great literature" rather than the simple existence of the literature. The same is probably true of science's terminal value: Bayes Theorem written on a rock in Andromeda has no value, but the clarity and knowledge and enjoyment of science that people who know the theorem get does.

Sure; but then why does it matter whether you yourself make the literature? Shouldn't you be even more fulfilled by becoming an investment banker, and giving money to support a dozen novelists?

Yet I don't know any novelists who would enjoy that.

That's cause there's a difference between reason and emotions?

I would feel a greater sense of fulfillment giving a kid an ice cream cone and seeing his eyes light up, than I would donating $1,000 to a a children's charity and knowing it would help dozens of children in need. I'm not saying that's better, just that my pre-programmed emotional responses react more strongly to it.

Your question seems to assume becoming a novelist is fundamentally a charitable act, or that novelists are working to maximize total world utility (unlike...everyone else?). I think most of them are just people who really, really like writing. I remember one novelist who said that the only valid excuse for becoming a writer was that you can't keep yourself from writing even if you try.

Even rejecting all that, the world does need a certain number of novelists, and not every novelist could succeed as an investment banker. If you assess yourself as having a certain high amount of literary skill and no even higher amounts of skill in other areas, then it may be that your best bet for contributing value to the world is in writing. Shakespeare's probably was.

Maybe because their comparative advantage is in novels rather than in investments?

That sounds like a fully general argument; for example, replace “make the literature” with “cure people” and “novelists” with “doctors”.

"All you can do in science is discover something before anyone else discovers it"

Mulling this over - maybe I'm taking a false view, but although I never had a particular admiration for the 'race to the south pole' type of exploration, the more general 'Going where no one has gone before' I do admire.

Because the second one does indeed do something - it establishes a new baseline that the next generation can start from. And so with Science - Newton described the real world with a precision greater than anyone before him, but off to the side Riemann established a new mathematics, which obviously had nothing to do with the real world, except of course Einstein proved it actually described the real world even better than Newton. And then of course the Quantum Theorist's described a 'Real World' greater still.

I've no particular advocacy for 'celebrity' science that races to be the first to something we already know can be done, although assuredly the innovation fostered by friendly or unfriendly rivalry has it's place, but science that actually expands our boundaries and tells us of new and different possibilities?

If I had the good fortune to be remembered for nothing more than have expanded the realms of possibility by setting up a base camp in unexplored science territory so someone smarter than I can mark 'Here be Dragons' a bit further out on the map, I could live with that - .


No one seems to have picked up on what I see as the main point of this post, so let me rephrase it in a Yudkowskian way:

Do you expect a Bayesian master to place intrinsic value on knowledge? If so, why?

Jeffreyssai sure as hell does. Does that answer your question?

Not the "why" part. And I'd rather know your opinion than Jeffreyssai's.

In that case, I guess the answer would be "no, but I expect them to place the same sort of value on knowledge as they do on art (in addition to the obvious instrumental value of knowledge)". I like reading new proofs of the Pythagorean Theorem in much the same way I like seeing new artwork by Sandro Del Prete. I'm not sure if I'm agreeing or disagreeing with you here - I think I'm mostly saying that I'm not really sure what "intrinsic" value means.

An elegant proof of the Riemman Hypothesis that forms spontaneously in the Andromeda Galaxy has no value to me, any more than a sequel to Mostly Harmless that forms spontaneously in the Andromeda Galaxy, but either would have value if there were people around to enjoy them

Actually, I feel the best response might be "taboo 'instrinsic' and 'instrumental'".

We are getting taboo-happy here.

The distinction between intrinsic and instrumental values is crucial in moral reasoning. The fact that we don't seem to be able to distinguish between them is a very big problem for moral reasoning.

I think empirical knowledge has intrinsic value. This is not because I'm a rationalist; I'm not a rationalist in the traditional sense (I don't think norms of instrumental reasoning are basic). Empirical knowledge has intrinsic value because it's cumulative. I consider this essentially an issue of identity - i.e., something that is cumulative is valuable. That's my definition of value. It's quite easy to show that something that is not cumulative has no value (most people agree that fads and repetitions are inherently without value) and that misattributions of value usually involve the misidentification of an endeavor as cumulative. It's harder to demonstrate an identity between being cumulative and having value though. There's also the issue of defining "cumulative" more clearly: collecting stones is cumulative in a sense, we can collect more and more stones, but is obviously not cumulative in the same way that science is. Science makes progress and this progress, I think, must sit apart from any supposed instrumental value - i.e., there's a sense in which science is not like collecting stones that doesn't involve reference to what science can do for us (doesn't make reference to any outside source of value).

As I've argued elsewhere, it's common to misidentify happiness and suffering as cumulative, and to then misattribute value to the alleviation of suffering or the promotion of happiness. This is a basic misconception of what a mental or emotional state is; if you have 100 happy people you can't say you've accumulated a lot of happiness any more than you can say you've accumulated a lot of red if you have 100 red balls. What you have is 100 happy people and not 100 times the happiness of a single person. Likewise, a person cannot accumulate happiness over their lifetime; a very happy life doesn't cause greater happiness at the nth instant than a moderately happy life. Emotional states are not cumulative and cannot (on my account) have value.

This is true of art too. There is, of course, a technological side of art that is cumulative; the development of perspective, of materials and pigments, and the development of photography and optics, these are all good examples in the visual arts. The development of music led to developments in acoustics. Art may be cumulative in small degree: artists avoid copying other artists. This, in fact, is probably what drives creativity in art: niche creation. The artist wants to strike out on his own and find a place for himself in the art world. But I think it's obvious this isn't cumulative in the same way science is; it's more like collecting stones than doing science. I do not, therefore, believe that art has any inherent value. (It may have indirect value by entertaining us, and thus creating an environment in which we can flourish in our cumulative endeavors, and by inspiring us. I think the latter is quite important. I think, for example, that fictive and fantastical and even erroneous concepts can be as important in inspiring us to real world discovery as logical or rational concepts, perhaps more so, and this is one of the reasons I do not consider myself a traditional rationalist.)

I should note that I don't believe I personally need to be involved in making scientific advances (although I have chosen to be). A person who is convinced of my ideas might take up political advocacy, or wealth creation with the goal of increasing the efficiency of others who are involved in the creation of knowledge, or might become an artist for the reasons I have given, or might just decide to become the best damn barista Starbucks has ever known. Knowledge creation requires an entire functioning, flourishing human society.


"It doesn't justify giving knowledge intrinsic value"

  • the whole point about things that you think are intrinsically valuable is that they require no justification. If I like having lots of knowledge or being the first to discover piece of knowledge X, I can say "intrinsically valuable" as a stop sign; you're not allowed to question that any more. Nice link to the enders game piece, btw.

Yes; so the way I phrased it was wrong. I will edit now.

Hrm... I personally am curious about stuff. I have a "I want to know... just because." for various things. A clever and elegant mathematical trick/derivation/proof "feels like" a good joke, in the sense that I seem to experience similar types of pleasure from both, in some ways. At least sometimes.

So for me, personally, it'd be a terminal value in addition to being an instrumental value. It certainly isn't my ONLY (apparently) terminal value, but it's one of them.

I don't think most people -- rationalists included -- really assign high intrinsic value to knowledge or science, even if they think at first that they do. They are valued instrumentally, for what they make possible.

If it were possible to have knowledge that was not available to consciousness, that you knew would never again be available to consciousness, but that would be stored away immutably in some part of your brain, would this have value? I don't think so. Knowledge that could never be used (in any sense), knowledge from which there could never be any benefit, that could be connected to other knowledge, would be valueless. If you say that such a thing would not be knowledge, and that knowledge by definition must be able to be brought to awareness, does that not hint that the value of knowledge is derived from the bringing to awareness and using for some purpose (if only aesthetic purposes)?

Likewise, art does not have high intrinsic value. Art is valued for its effects, for the quasi-mystical feelings it can inspire and the joy it can bring. If Bach's music did not move me as it does, if it did not inspire me, the only value I would assign to it is value based on whether and how much it moves and inspires other people.

The instrumental value of science is that scientific progress or lack thereof is what will make the difference between a cosmos of life and mind, versus a cosmos of dead matter in which mind was a transient blip on a single dust speck. It's not that it matters which scientist gets there first - it's that it matters whether we get far enough within the time we have.

The time we have is primarily limited by the same scientific progress, so what matters is sanity of what we do, not speed.

Speed matters somewhat, in that we might overpopulate, or use up fossil fuel or helium or other resources, in a way so that the difficulty of escaping Earth continues to be just beyond us.

But this is all talking of instrumental value. I don't want to talk about instrumental value here.

Are you so sure that good art isn't destructive? It makes the rest of the world seem bland in comparison. When I started reading, at Eliezer's recommendation, the classic work of literature Fate/Stay Night, I found I had trouble reading other, lesser books, because, who cares?

An analogy might be to the Nymphs of Dungeons and Dragons - people who gazed upon their beauty would go blind or even die from despair that they would never see something so beautiful again.

By induction, never do anything well.

See Eliezer's postings on Superstimuli in OB. And more specifically his quote from the babykiller/superhappies series: "It's bad enough comparing yourself to Isaac Newton without comparing yourself to Kimball Kinnison."

And which would you rather be - a great scientist, or a great artist of some type? (Pretend that great scientists and great artists are equally well-paid and sexually attractive.)

In a world where war, death and suffering are dealt with for good, I'd be an artist (Edit: who, nevertheless, is intrinsically interested in some scientific problems such as abiogenesis, origin of sexual reproduction, life based on non-DNA replication, Fermi paradox, simulation hypothesis etc.)

In today's world, I'd be a scientist -- even if scientists had less money and sexiness than artists.

Edit: Haven't noticed this: "Also, ignore the possibility that your scientific work can make a safe Singularity. That would be science as instrumental value. I'm asking about science vs. art as intrinsic values." In my reply above, I chose science because it can fix the world, so it looks like I was going for its instrumental value.

All you can do in science is discover something before someone else discovers it.

To my mind, time-discounting feels like a major reason for doing science. It's better for humanity to discover stuff earlier rather than later. For a particularly clear-cut example, there's obvious instrumental value in discovering penicillin 20 years earlier.


I saved this to "drafts". Didn't mean to post it. Does "Save to Drafts" just not work?

It hasn't been posted. Check "New". You won't see it. But you, and moderators, can see drafts.

I do see it under "New".

"What do the rest of you think? Is there a strong correlation between rationalism, giving knowledge high intrinsic value, and giving art low intrinsic value? If so, why? And which would you rather be - a great scientist, or a great artist of some type?"

I've read through many of the posts on this site, but this is the first I've answered--the first I've felt knowledgeable enough to answer.

Generally speaking there might be a strong correlation--I've seen that pattern often (I'm currently in medical school, and in undergrad was the only LibAr in a houseful of engineers). There is, especially in the West, a dichotomy drawn very early and very deeply between imagination and knowledge--a belief that somehow they must be mutually exclusive. Even Einstein (who really ought to've known better, since his imagination was fundamental to his attainment of knowledge) emphasized the distinction ("Imagination is more important than knowledge"). Most forms of higher learning demand imagination and discipline; artistic creativity demands them as well, albeit in a different tenor.

I'm an artist, and my high valuation of knowledge (and growing valuation of rationality as a tool with which I can temper it) interlocks seamlessly with my desire to create. A greater understanding and integration of the world, more information, more finely honed perception--all of those things increase the quality of both my life and the things I create. This is as true of the areas of physics and mathematics and cellular biology as it is of English literature. Greater understanding produces better art.

It's doubtful that I'm a singular, or even rare, instance of such thinking; many of the Renaissance masters would probably agree. I've often wondered what Michelangelo might have produced if he'd somehow been able to share in Mr. Yudkowsky's beautifully elaborated HPMOR vision of humanity triumphantly ascendant over death and mourning; I suspect it would have outshone his 'David'. In fact, Mr. Yudkowsky's a great example of someone who feels no need to choose between artistic achievement and profound knowledge.

As to your question of preference...I can't choose. If I were offered one of the two, I'd seize each as eagerly as the other. And--like Goethe, Franklin, Khayyam, da Vinci, Pascal--if offered both...I'd choose them both.

Although this is a topic in which I am quite interested (given my own background), I am downvoting this post on account of the following needless provocation:

20th-century developments in orchestral music killed orchestral music

If you're going to casually toss out something like this, the least you can do is be clear about what you mean. The last time I checked, orchestral music was still being written and played. (Three guesses how I might know this.)

Actually, there are ways of interpreting that statement in such a way that I would agree with it. Based on previous comments of yours, however, I doubt that's what you intended.

No, I think I meant just what you think I did. Orchestral music is dead. Name one great composer since Stravinsky. With the money we spend on music, and the number of composers trained, we could have had a dozen, maybe a hundred, Beethovens since Stravinsky. What do we have? John Williams and Danny Elfman.

There may be great orchestral composers out there somewhere, but the orchestral music scene is too dead to find them.

Name one great [orchestral] composer since Stravinsky

Are you serious?

Just restricting ourselves to the cream of the crop:

Boulez, Carter, Babbitt, Sessions, Berio, Nono, Ligeti,...

All these and many more have written important works for the orchestral medium, and all were born after Stravinsky...indeed, all of the above list except Sessions were born in the 20th century.

I really, honestly, don't want to be rude or confrontational, but...

...the fact that you would cite John Williams and Danny Elfman (both of whom work in the film/television industry) as your idea of the contemporary orchestral composer shows how completely uninformed you are. It's as if there's a whole entire field of human activity of whose existence you are entirely unaware.

Now, no one should be expected to know about everything. But Eliezer's point about science ( generalizes: there's a whole, wide world of stuff out there. Yes, art music has a PR problem...but then, so does science, at least among the general public. I presume people here, including yourself, know better in the case of science, so what gives?

I ask the following not (merely) as a rhetorical question, but out of a genuine desire for insight: what on Earth gave you the impression that you were in a position to judge the state of contemporary orchestral music?

There may be great orchestral composers out there somewhere, but the orchestral music scene is too dead to find them.

Just in the past few years, James Levine at the Boston Symphony has been commissioning and premiering a whole series of new orchestral works by living American composers, including Carter and Babbitt. Again, just to take an elite example.

Appreciate this. I've posted to similar effect on my nascent blog, The Grouchy Musicologist:

Indeed, the aggravating nature of the offhanded comment you responded to was what got me off my butt to create said blog in the first place -- there'll be a lot more where that comes from, though. The same standards of rationality that we ask of everything else can apply to discourse surrounding music, even though that seems (incredibly) to be non-obvious to many otherwise very rational people.

Well thank you Mr. Bardic Conspiracy. You really think yours is the only Conspiracy in the world besides the beisutsukai?


Do you think a Bayesian Master necessarily places a high intrinsic value on knowledge? If so, why?


What happened to wisdom? Knowledge with no understanding is useless.