Comment by bairyn on 2-Place and 1-Place Words · 2012-12-10T07:53:03.473Z · LW · GW

The vocabulary of the constructed language Lojban seems, to me, to be, overall, much more useful; this seems to result naturally from the language's simplicity and lack of irregularity and arbitrary restrictions, which make it difficult to naturally think of verbs in terms of multiple arguments. selbri, Lojban's equivalent of verbs, regularly make direct use of more than one, and even two, arguments. For example, {zdile} (note: the Lojban community uses the convention of using braces to quote Lojban text inside English text) is the translation of "fun". Its definition is "x1 (abstract) is amusing/entertaining to x2 in property/aspect x3; x3 is what amuses x2 about x1."

It's difficult to construe, in a way that's rigorous to at least a useful degree, any usefully meaningful interpretation of "fun" without any notion of a subject or experiencer in contexts in which it commonly occurs, whether the result is based on a fixed subject or not. (It is this reason, by the way, for which I hope that the fun sequence, which I haven't yet read, makes it really clear that our utility functions may not necessarily preserve our system of "fun"; it's a property of us, not of the universe, so we may find another way to experience "fun" more valuable; it' s entirely the result of our own minds. An essay which goes on in great detail about similar ideas is the one at, which, on a personal note, largely represents my cur rent views, with the exception that, while "happiness" is likely-to-me one of the most important factors in our utility function, it is not our only value (I don't remember whether this was laid out by David Pearce); regardless, eliminating suffering is extremely important to me. (Any conception in which our universe, or mathematics in general, is somehow fundamentally associated with an absolute moral, value, or goal structure, seems, to me, to be the result of confusion.) The essay had a dramatic emotional impact on me when I read it three years ago (I was 15 at the time, had never heard of Less Wrong, and had not yet conceived of the abstract notions of systemic cognitive processes that reliably increase the accuracy of beliefs and reality and of optimizing for utility; I still believe that the principles I remember from David Pearce's writings remain consistent with what I've learned from Less Wrong; I was, in fact pleasently surprised to find a link to them on the LW wiki article of the fun theory sequence as an external link.).)

Another example is that of the very concept discussed in this essay, that of "beauty". The closest single-world translation in Lojban (I'm referring to the dialect defined by the CLL and the updates formally accepted by the BPFK) is {melbi}, whose definition is "x1 (abstract) is amusing/entertaining to x2 in property/aspect x3; x3 is what amuses x2 about x1.". In order for this concept, by this definition (which happens to be meaningful and correspond very closely to our own, of "beauty"), to be meaningful, there needs to be a subjective e xperiencer, the second argument to the function {melbi}.

Lojban is far from being any supposed "perfect" language; it's the result of arbitrary principles and unpredictable complexity. Still, it's based on really useful principles; this is why I like it much more than I like any other language of which I'm aware. I am, however, only slightly hesitant to commit to learning it, which requires an immense investment of time, for the same reaso ns I'm hesitant to commit to learning and using a language that regularly appeals to practicality to the extent that Haskell does; there are many ways to approach the problem, some of which are dramatically worse, by some goal structures, than others, and none of which are based on a perfect system, one that is consistent and complete. As a language for humans, though, Lojban is reall y nice and well designed, often even in areas that aren't necessitated by its principles (its vocabulary, particularly its definitions, for example).

I suspect that, while putting serious effort into learning Lojban, for most people, enables them to think in a slightly more rigorous way, using the system of Lojban; learning, say, Agda will be useful to a much greater extent. Unfortunately, very few people would be capable of and motivated to do the latter, which is p art of why I think it'd be nice if Lojban were thought of as important to learn as English.

(By the way, I've also been happy to see rigorous distinction between quotation and referent briefly mentioned in several places around Less Wrong. In my experience, learning Lojban has helped members of the community learn this distinction if they hadn't already. In Lojban, there is no ambiguity, in neither writing nor speech, between a quote and a referent; you would sound quite si lly if you said {la bairyn cu se cmene mi}, rather than {zo bairyn cu se cmene mi}, as Byron is my name; the notion of Byron the person being my name is quite silly indeed, and the difference is quite obvious in Lojban, the arbitrarily designed human language. If a person hasn't internalized the distinction, I suspect Lojban would be likely to help them do so, whether they deliberatel y attempt to or not. I would also like to point out that there are often many ways to dereference a quotation; obvious examples include interpreting a word and converting a variable, such as {k o'a}, to the sumti, the latter of which depends on the environment.)

While it'd be trivial to arbitrarily define a Lojban selbri in a rather meaningless way, it seems that the designers were careful to construct only rigorously meaningful gismu. I never (thought much)[] about a rigorous meaning of "should" until I tried to translate it while speaking in Lojban; it turned out that this had been a problem faced by many speakers, many of whom just gave up, it seems.

Meditation: Some time after that, I finally understood how I could more formally understand how I interpreted "should", which was, quite simply,

... ... ...

Reply: "optimal for some goal structure", which, of course, depends on the goal structure in question; our own human values is an obvious implicit x2. (I haven't yet read the metaethics sequenc e, but I expect it contains essays describing topics similar to that of this discovery). Perhaps translating into Lojban is often a good strategy for Tabooing ideas. (Thanks, Eliezer, for the meditation idea!)

Comment by bairyn on 2012 Less Wrong Census/Survey · 2012-11-04T23:57:18.312Z · LW · GW

Took it. All of it.