I have standardized on giving everyone exactly one clap — equivalent to a recommend. I want to avoid contributing to runaway applause inflation (the way that ratings inflation makes rankings on ebay completely worthless).
If an article isn’t in some way exceptional (in other words, if it wouldn’t make its way into my recommendations under the old system) it gets no applause. After all, my recommendations show up both on my profile page and automatically tweeted & sent to other social networks — so, out of respect for my followers there, I refuse to spam them with articles that are merely adequate. Adequate gets no claps from me.
The mechanics of how this turns into actual payment is unclear, so the way that my one-clap policy affects payment is unclear. The sensible thing seems to be to normalize by actual clap distribution per user: in other words, every clap a user gives over a month is worth the portion of their subscription fee dedicated to payment divided by the total number of claps they gave out (so my one-clap policy is equivalent to your five-clap policy). It may be that they’re doing a slightly easier but less sensible strategy, wherein the total pool is evenly divided by the total number of claps without normalization by user — which would only produce sensible results if applause behavior was comparable between users.
As a writer — I haven’t been able to determine whether the number on my stats page representing applause is the number of unique recommenders or the actual amount of applause, for new articles. (Maybe everybody is giving me one clap? Maybe they changed the label but forgot to update the variable it was pointing to? Maybe because the article was drafted before the transition to applause but published after it the old version was grandfathered in? Or maybe you need to actually visit the story to get the ‘real’ applause count?) Anyway, calculating value-per-clap isn’t possible without more details about the pricing algorithm. Ev Williams has written elsewhere that predicting the value of an article isn’t possible; this probably still applies if you know the number of claps.
Claps clearly still work on the old recommendation system, hence the lack of negative claps. (The ability to un-recommend is gone unfortunately.)
Ev said that during the last billing period (so, I guess the tail end of August) 75% of authors with members-only stories got paid something and the range was between ~$3 and ~$300. 75% sounds like it might be the percentage of members-only stories with a non-zero number of claps, which would mean that $3 might have ended up being the value of one clap for the tail end of August. If I recall correctly, the period during which the applause system existed wasn’t a full month, so the sample size would be smaller than a full month and the prices higher (since the entire portion of the subscription fee dedicated to paying for stories directly was allocated after claps were implemented, right? Or, were recommends used to allocate funds previously?)