Just when you thought this thread was dead.....
1) Thanks to all involved for your hard work combining and organizing the various versions of a release on Stern pins of the past almost decade. It's definitely an improvement. Look forward to seeing the results for other manufacturers. It makes reading about, researching, comparing, selling, and buying machines easier.
2) I agree with Damonator and still think game groups should be ranked by weighted average (as in the example by robin in post #74 (https://pinside.com/pinball/forum/topic/operation-premium-le/page/2#post-7421319). However, I do understand the arguments not to do that.
3) Suggestion: I would *LOVE* to see the custom tab have an easily accessible method to show the ranking by game groups with a weighted average for the entire group. The default custom ranking tab representation of "count every version of a machine as an individual game and spit them out" is somewhat interesting, but not very useful to me because of all the "duplicates" clogging up the rankings.
4) I think it is weird and bad form for the main tab of the top 100 rankings to:
a) list by game groups
b) rank by highest average rating for a particular "trim" within the game group
and yet
c) list the *TOTAL* number of rankings in the short summary that is seen without expanding the game group.
I would suggest not including the number of ratings in the default view (which is the unexpanded listing). It seems misleading from both representational and statistical points of view. I can glance at the list, and see that the rating numbers decrease from highest to lower values. I would submit that, like my assumption, most people's assumption would be that the rating displayed in the default view is generated by an average that includes the total number of ratings listed in the default view. That is exactly what I assumed until I started expanding some game groups 4.5 years ago when I started reading Pinside, and figured it out. But I was surprised by the result. My assumption was that Pinside would treat the statistics "correctly" the way most would assume based on the representation. I would also guess that *most* people looking at the default top 100 list don't understand the nuances.
If you must include a number of ratings in the default view, make it the number of rankings that corresponds to the "trim" that is giving the rating number represented in the default view. It really doesn't help to show a big number ("hey, everyone likes that title") and find out that the rating is really generated by a subset (and sometimes a non-majority or even a small subset) of the total game rankings.
I know that the top 100 isn't meant to be a definitive ranking and it has flaws and can be manipulated to some degree. However, showing the ranking score right next to the total number of ratings is really very confusing and bad form, IMO.
....shuffles to the back of the room again...