Curation or Algorithm? … David Hume weighs in.

I’m a philosophy nerd. I read quite a bit of the stuff, I listen to podcasts, and I even take online courses from time to time. This is no boast, and I’m no dilettante (those of you who know me get this), but I do dig the stuff and furthermore, I’m convinced that all of us who work in this topsy turvy, madcap, tech circus could periodically benefit from the wisdom of the greats to enlighten us with perspective on the human condition – - you know — “user behavior”. Recently, I’ve been following the debate/ discussion that can summarized as ”Man vs. Algorithm” with regards to the filtering and recommendation challenge for digital content.

The “recommendation problem” has been a constant theme in my career and, indeed, its a big part of the internet’s story. How do you filter all of that information so that a single human can acquire the result they desire? “Curation” has been a concept that’s been floated as, if not a sufficient solution, then at least a very necessary component of any attempt to solve this intractable problem. For the purposes of this post, when I say “curation”, I mean a process that involves human editors/curators — I realize that technically, “curation” can be accomplished by a computer, but this over-complicates things, and the really interesting discussion is the whole “John Henry vs. the steam drill” (or… curation vs. algorithm) thing.

Its an even more interesting problem if you perform the following thought experiment: ok, so let’s say curation is the answer – well, what does the PERFECT curation experience look like? Those of us who build stuff often start with these conditions (knowing, sadly, that we will never achieve it) as a means to define the parameters of the possible outcomes.  Turns out, as usual, this debate has been going on, in one form or another, for quite a while, and fundamentally, the issue goes much deeper into human nature… so, David Hume – you’re a smart dude, what say ye?

I’m a big fan of David Hume, and “A Treatise of Human Nature” is, IMHO, one of the greatest practical philosophical works of all time. (I actually think he is the patron philosopher of product and design even more so than Karl Popper, but that’s for another post).  So, I was recently super stoked (see, no dilettante here) when I was educated about a Hume essay entitled “Of the Standard of Taste” while listening to this episode on the Philosophy Bites podcast (which I highly recommend). It turns out that the work is actually quite seminal in the field of aesthetics, while for Hume, it was slap-dash 34 paragraphs he wrote quickly to make up a gap in a book of essays – - must be nice when your throwaway stuff is makes you the poster boy for an entirely new field of study (Hume called it “criticism”, and only later is a area of study labeled “aesthetics”).

The podcast episode provides a really great analysis of the essay, as does this entry from the Stanford Encyclopedia Philosophy. In the essay, Hume defines how beauty in art is defined, and boldly, he proclaims that beauty (you can also read “quality”) is not subjective — that is, not “in the eye of the beholder” at all. For those of you who are as lazy as i am, here is a 5 bullet point breakdown:

  • Hume, the quintessential empiricist, believes that all general rules of art are based on experience, not on a priori knowledge (in opposition to Kant and subjectivism). This means that we derive our sense of beauty and quality through our senses, and it is not innate in us.
  • Questions of beauty and whether a work is “great” or not IS a matter of feeling and pleasure, BUT, that doesn’t mean that there isn’t a “right answer” as to whether something is beautiful or great. He describes it as nothing less than “fact” in this passage:

But if we consider the matter aright, these are questions of fact, not of sentiment. Whether any particular person be endowed with good sense and a delicate imagination, free from prejudice, may often be the subject of dispute, and be liable to great discussion and enquiry: but that such a character is valuable and estimable will be agreed in by all mankind. Where these doubts occur, men can do no more than in other disputable questions, which are submitted to the understanding: They must produce the best arguments, that their invention suggests to them; they must acknowledge a true and decisive standard to exist somewhere, to wit, real existence and matter of fact; and they must have indulgence to such as differ from them in their appeals to this standard.

  • Taste is varied, just like moral sense, but as with morality, some things are simply “better” than others just as some acts are accepted as “moral” or “immoral” within human society.

It is indeed obvious, that writers of all nations and all ages concur in applauding justice, humanity, magnanimity, prudence, veracity; and in blaming the opposite qualities. Even poets and other authors, whose compositions are chiefly calculated to please the imagination, are yet found, from HOMER down to FENELON, to inculcate the same moral precepts, and to bestow their applause and blame on the same virtues and vices.

  • Therefore, we must have a “standard of taste”, just as we have our morality codified in laws.

It is natural for us to seek a Standard of Taste; a rule, by which the various sentiments of men may be reconciled; at least, a decision, afforded, confirming one sentiment, and condemning another.

  • So… who/what defines the “standard of taste”? He has an answer for that, as well, and the answer comes straight from the mouths and pens of “true judges”.  In fact the “standard of taste” is specifically defined as the “joint verdict of the true judges”. He goes on to define how one becomes a “true judge” and how we might identify these chosen savants.

Though men of delicate taste be rare, they are easily to be distinguished in society, by the soundness of their understanding and the superiority of their faculties above the rest of mankind. The ascendant, which they acquire, gives a prevalence to that lively approbation, with which they receive any productions of genius, and renders it generally predominant.

Ok, I’ll break here for a moment just to comment, because to me, this is just great stuff and fodder for many a bar-room argument… there IS a greatest rock album, and it can be legitimately argued. In Hume’s world, music critics (at least the PERFECT ones) aren’t bitter and/or obsequious leaches “dancing about architecture”, rather they are exalted leaders, men of honor, and important social glue for society.

There is a bit of a problem (and some controversy) in his theory, because Hume’s qualifications for a “true judge” are stringent bordering on the impossible – - some philosophers say that, in fact, Hume was really referring to an ideal and that he did not believe that such “true judges” could exist, but we’ll stick to what we can directly infer from the text. I’ll bullet down his qualifications below:

Whether any particular person be endowed with good sense and a delicate imagination, free from prejudice, may often be the subject of dispute, and be liable to great discussion and enquiry: but that such a character is valuable and estimable will be agreed in by all mankind.

  • By “good sense” (he also refers to this as “strong sense” elsewhere in the text), he means that the person needs to intelligent, reasonable, and able to understand the various elements within the subject matter he/she is endeavoring to criticize. The judge must also be able to use their intelligence and good sense to organize their thinking and argue well – - they have to defend their position, because they are “right”. So, for Hume, you CAN lawyer your way into proving that the Beatles are the best band rock band ever… or whatever.
  • Owning a “delicate imagination” (also referred to as “delicate sentiment”) means being very in tune to the subtle nuances and details of an art form. Hume is adamant that in order to empirically acquire this quality, one requires a lot of training, practice, and experience in order to nurture the judgement to make the fine-grain comparisons between works and their components.
  • Finally, one must be “free of all prejudice”. For Hume, this primarily meant that one must be able to rise above any bias derived from allegiance to nation, religion, race, or culture. Yeah, no problem.

So what do we have here, aside from the most pretentious job posting of all time for an editor? I think we find some reference material for framing our discussion about “curation vs. algorithm” and the recommendation challenge. Clearly, if Hume were pressed to answer the ”Man vs. Algorithm” question, he would come down squarely on the “man” side, and as it happens, he has a really nice essay to back it up. But deeper than that is his belief that there IS an objective answer — that we can KNOW whether art is good or not.

It’s curious to me that the “joint verdict of the true judges” smacks a little bit of what we might today call crowd-sourcing.  Certainly Stack Overflow and Wikipedia have pretty tried and true mechanisms for separating the “true” from the “untrue” judge – - and when we reach that “joint verdict”, it is sort of starting to feel like pornography – we know it when we see it. What would Hume think of a recommendation engine that delivered subjectively good results, but objectively bad ones… that is – what if you have “bad taste” and Spotify, Beats, or Pandora just keep feeding you more “bad” music? I think Hume would think that is a bad outcome and a useless endeavor… probably even immoral. I think he may argue that your continued spiral into crappy content just makes you and society dumber and worse citizens…. hard to say, but probably not a bad guess.

Certainly, the qualities of a “true judge” would be hard to replicate with computer code — it is hard to think of a programmatic way (short of science fiction-y AI scenarios) to understand the nuance and context associated with a “delicate imagination”. That being said, a computer is going to be a heck of a lot better at “absence of bias” than a typical human, so maybe there is a Humeian curator algo out there yet. The interesting observation for me is that the standard “product to product” collaborative filtering engine – i.e. Amazon – has elements of objective and subjective recommendation. The collective action of millions certainly isn’t Hume’s idea of the ideal “true judge”, but just maybe the result is the same with enough events?

So who would be the patron philosopher of the the “uncuration” … the “give the people what they want and not what they deserve” camp?  Well, I think we look at the opposing positions of the time regarding aesthetics — Hume was an empiricist, and his treatise on taste came from an empiricists viewpoint. So, I would go to two philosophers who stood opposed to Hume back in his day. Immanual Kant, because although he wasn’t, by most accounts, really a subjectivist, he basically spent his whole career and his greatest work (The Critique of Pure Reason) directly arguing with Hume and his empiricist views. His direct work in aesthetics was Observations on the Feeling of the Beautiful and Sublime. I would also look to Søren Kierkegaard, who is basically the poster boy for subjectivism. Both philosophers believed strongly that beauty and quality in art are COMPLETELY in the eye of the beholder, and they are derived from nothing but pleasure and feeling. In their ideal internet, feeding you more crap on top of the crap you already consume is a perfect goal, as long as enjoy it – - but I think we all doubt that path reaches any decent goal and therein lies the rub…

As human beings, we have subtle tastes, a huge dependence on context, and, as Hume would say, “delicate imagination” when it comes to the stuff we love. I love me some bad action movies… i mean some really bad ones… but I don’t love ALL bad action movies, and in fact, I can’t even describe to you what makes me love the ones I love and hate the ones I hate.  Some of my friends understand my choice immediately, and it is because they are free of bias and share my “delicate imagination” which enables me to know that Predator is totally awesome, while The Expendables mostly sucks (again – not a dilettante).

I don’t think the real debate is “Curator vs. Algorithm” — I think that debate exists and is relevant at the tactical level, but I don’t think we’ve yet decided, collectively, what the goal of such mechanisms are supposed to be.  When we design these things, we have goals in mind and we know the technical and semantic limitations of how far they can go, but I know from experience that even the best-case scenarios fall pretty short of what we would like.  I do think that its useful, from time to time, to get our heads out of of the solutions based purely on discussions UX and data science and remember that people who were a hell of a lot smarter than us who had no internet still have important ideas to convey… i’ll leave you with this.

 Many men, when left to themselves, have but a faint and dubious perception of beauty, who yet are capable of relishing any fine stroke, which is pointed out to them. Every convert to the admiration of the real poet or orator is the cause of some new conversion. And though prejudices may prevail for a time, they never unite in celebrating any rival to the true genius, but yield at last to the force of nature and just sentiment. Thus, though a civilized nation may easily be mistaken in the choice of their admired philosopher, they never have been found long to err, in their affection for a favorite epic or tragic author.