I drafted the words with the phonology rules of http://selpahi.de/ToaqAlphaPrimer.html caiq is the first syllable of the word and ce the second.

But I grant you that at the moment I don't understand enough about phonology to publish a working draft of a language. My intent with this post was more to present the compounding system that I consider to be useful.

Phrases tend to take on meaning distinct from or supplementary to their constituents. In written English, you can highlight this with quotes or capitalization, both of which are disruptive and therefore under-used. In spoken English, marking a special phrase is very awkward, and is done extremely rarely.

It'd be nice to have a function word or pair of function words for this. It could probably double as a parenthesization mechanism.

Please forgive me a bit for mixing different ideas over multiple post in this thread with a bit of overlap. I consider the ability of a language to specify relationships very valuable and underdeveloped in English. Latin has a word for mother of father. English has only grandfather or grandmother. It has ugly constructions like great-grandfather.

In my draft I have the following root words:
ba = 0
ce = 1
di = 2
ma* = female
ne* = male
caiq = parent

Out of those roots I can create: caiqma = mother
caiqne = father
caiqce = grandparent
caiqcemaba = grandparent (parent of the mother)
caiqceneba = grandparent (parent of the father)
caiqcemace = grandmother
caiqcenece = grandfather
caiqcemana = grandfather (father of the mother)
caiqdi = great-grandparent

This way of specifying relationships is quite efficient. In case you want to distinguish your parents not by gender but by which parent is older and which is younger, you can simply use the syllable for "younger" instead of the on for "female". That way the language can translate easily from languages that have different words for older and younger brothers, while not forcing lanugage users that don't want to make distictions based on gender or age.

Why four letters for caiq? Because it's based on cai with simply points to the parent node in any graph. Combing cai with the sylable for knowledge from authorities fwe, caifwe becomes teacher. It's easily extensible so that caifwece is the teacher of my teacher. English has no word for teacher of my teacher and my language can still do it in 8 letters. It can even do teacher of the teacher of my teacher in 8 letters a case where English feels like Pirahã.

Do other words for family relationships are:
fuiq = sibling
caiqfuiq = aunt/uncle (parent's sibling)

Out of that a person with the same teacher as me (classmate) becomes from the structure we already have fuifwe. We get a new word of caifuifwe with means a person with whom your teacher learned together under his teacher. We get that word without the language learner having to learn it explicetly.

There will be many cases where more complex relationships can be easily expressed with that system. Via Sapir-Whorf I would expect that this well structured system of relationships makes it easier to think about more complex relationships.

*ma/ne : Those are very provisional. Likely it's no good idea to have two nasal consonants at this place but instead use two consonants that differ more from each other to reduce the cognitive effort that's required to hear whether someone says one or the other.

How about affixes for central vs noncentral usage of concepts? For example, you might make a rule that you can stick -cen on any content word to mean something is a central usage of the word and its meaning has not been stretched, or -sep to mean that it's figurative or the meaning has been stretched or there's a caveat.

Can you add a way to be notified if at least two people are currently joined?

I've just made a few updates to the online implementation. Specifically:

• There's an in-game chat.
• When waiting for games, there's an option to leave the tab in the background and get notifications when players join and when the game starts. So if other people aren't visiting at the exact same time, you have a better chance of getting to play.
• Private games don't auto-start at 6 players, you can have more if you want.
• Miscellaneous minor bug fixes.

I've got a few people interested in an effective altruism version of this, plus a small database of cards. Suggestions on how to proceed?

Spent about 20 minutes playing online, I have some technical notes and general impressions.

Technical (skip this if you're not Jimrandomh):

• The timer feels way too long, especially as people get to know the cards better and don't have to read all of them.
• When choosing card pairs they are displayed in long rows, so for 3 people someone's first and second cards are on different rows. That's very unintuitive. Maybe put the pairs in separated columns?
• When judging, seeing the timing of the cards coming out can skew the judgement, and also makes it easy to guess which card is the control.
• The website works smoothly, well done!

Here are my main takeaways:

• The cards are excellent, a lot of them are either very funny or are doing a good job explaining things quickly. For some, it's hard to tell which :)
• Unfortunately, the jokes that happen during play itself aren't funny at all compared to the cards. A lot of times there isn't a single card that will give a "funny" answer, am I supposed to choose the logically appropriate one instead, then? I wonder if I'd be more likely to buy the best cards as a poster than as a card game.

I'm going to try and invite some non-LW friends to play, see if they like it or run away screaming in confusion.