But isn't it all meaningless?

Didn't Wittgenstein say that private languages are impossible?

Some people argue against conlanging that Ludwig Wittgenstein, one of the 20th century's most important philosophers, concluded that 'private languages are impossible', which sounds a lot like 'conlangs don't work'. The argumentation, simply put (and as far as I understand it), goes like this: Wittgenstein observed that languages are the result of societal conventions regarding the meanings of words and the rules of grammar. From this, he concluded that private languages are impossible.

Hence, so the argumentation goes, 'conlangs don't work', and what conlangers do is just a lot of moonshine. This, however, is contradicted by solid facts: while artificial IALs like Esperanto have not yet found the success intended by their creators, they have shown to work; there are communities successfully communicating in them. (The same goes for popular artlangs such as Quenya or Klingon.) Or take a look at a conlang translation relay: while the meaning of the relay text indeed shifts from one translation to the next, and the final result often has little in common with the original text, each version closely resembles the preceding version, indicating that a successful transmission of textual information encoded in a conlang has taken place. It would be interesting to conduct such an experiment with real ethnic languages; most likely, the results would be similar. If texts written in conlangs were indeed inherently meaningless, one would expect that the relay text would be reduced to utter gibberish after the very first move - which is not the case.

So, where's the mistake? Was Wittgenstein wrong? It can hardly be doubted that languages are indeed based on social conventions. The meanings of the words and the rules of grammar are completely arbitrary; they are the way they are because people agree on them. The fault then evidently lies in the interpretation of the sentence 'private languages are impossible'.

The point is that conlangs aren't at all 'private languages' in the Wittgensteinian sense. What Wittgenstein meant with 'private language' is a language that is inherently private, i.e. a language that in principle can never be learned by anyone else than the person making it up; and he concludes that such languages cannot exist because whenever someone makes up a language, this language could in principle be learned by other people as well. The only difference between a natlang and a conlang is that in the former, the meanings and rules are the result of millennia of cultural evolution, while in the latter, they are set by a single individual (or a group of collaborators). Nevertheless, the meanings and rules of a conlang are a convention that others can share. Hence, 'private languages are impossible' doesn't mean 'conlangs don't work', but almost the exact opposite, namely, 'conlangs work because they aren't private'.