:shock:
"Romans_I_XVI" wrote:
Also on the Roku 3 it seems to be that it's putting all of the non-BrightScript (NDK or Marmalade) games towards the top.
Oh hello! - nice to see you around. Your guess of non-BrightScript games benefitting would be incorrect - see the screenshot above, out of 7 games at the end, 5 are non-B/S (the other 2 i don't know). If any trend, i rather seem to notice the
old games seem to top the chart - but then "Chess" and "Backgammon" (which are new) are listed high without being crazy popular.
"adamkaz" wrote:
There was an offhand mention in an article I read recently that Roku sorts the channel sections by the number of hours of content streamed on the channel - can't seem to find it now. This would explain some of the difficulty with sorting the games and apps sections - since streaming video isn't really what they do.
I wonder if someone at Roku implemented a random sorting for tie breaking (channels that play no streams), to prevent channels from becoming popular simply because of publish date or title chosen.
It makes sense to measure
user engagement to determine popularity - like say number of times app was started or total time spent using it. Heck,
Amazon Underground actually pays game developers for the time users spend in their game! - the more time spend, more $. No ads involved :shock:<gasp>. Measuring video play was probably a shortcut, an easy hack to exclude time spent in screensaver or channel abandoned for days in list-of-videos menu, doing nothing. But such metric makes no sense for non-streaming apps, none whatso<censored>ever.
Re the random sorting, Hanlon's razor likely applies - i.e. no malevolent or benevolent intent but rather their sort routine is unstable (e.g. quicksort/shellsort are unstable). But that's trivial to fix by enhancing the key, taking into account say user rating (some additional variable). The instability of the chart is a bit of "canary in the coal mine" - while not a big deal by itself it indicates bigger problems behind.
