inim said:Maneki-Mew said:I mean yeah, but what does your heart say?
My heart says that my ratings, plus minus some taste related variations, are very close to anidb.net's normalized ratings. Which means they are on average about 2 points below MAL's ratings. Which is perfectly fine, I maintain my list for myself and not to counterweight what I consider a broken rating algorithm on MAL. anidb.net displays both, rating and not normalized numerical averages, and MAL long term has no option to do the same. Otherwise, eternal rating arms race and inflation renders the scores completely meaningless.
You can use the two of us as examples. We have > 70% affinity, on over 220+ shared items, with an absolute rating difference of 1.9 over our shared set. If you look at the affinity formula, you'll see that it also has that calculation of the average and adapting absolute scores relative to your personal average before comparing. It basically compares the deltas of your ratings to your average rating and not the raw numbers. That means our lists are pretty similar, except for the fact they are shifted by 2 in absolute terms. You are free to use 7 as "average", but I personally prefer to use 5. And that's pretty much the only difference we have :)
As for bad shows, I don't really watch too many bad TV shows because like you say they take long. But I love trashy 20th century OVAs, and rating range 1 to 3 is pretty much where they are. MALGraph has a good plot in which you can see how much watchtime you spend in each of your rating classes. Me too spent most absolute time in 7s, disproportional much time in 8+s (because longer shows tend to get higher ratings from me), and very little time in my 1-3s. But I have many 1-3s because most are shorts and OVAs. Again, this makes perfect sense to me.
And a 6 is not "just fine". Only 326 of 798 (40.8%) anime on my list have a rating of 6 or better. On your list 706 of 1002 (70.2%) have ratings of 7 or better. Which means my 6 is "worth" more than your 7, relatively. And after normalization, only relative difference matters. Maybe I'm cursed because I have a college degree in science and not psychology, but to me that all feels very natural :)
I export my ratings to anilist and anidb, so I don't work around the lousy MAL formula. I deliver very good raw data in the sense that it is well distributed, and leave it to the servers to make the best of it. Some do better, some do worse - but why would I care?