14:37 yes, i mentioned something like that
14:37 My hunch is that question I would want is something like "how are you: bad - ok - good"
14:37 or maybe 'how's it going'
14:37 heh
14:37 and let fx take the blame for everything
14:38 that way there is always some 'up' and 'everything' affects down
14:38 so in my experiment to quantify questions... you'd get two at a time.
14:39 so one can calibrate them directly
14:39 (two of the candidate questions)
14:40 leo's satisfaction questions are quite straightforward
14:40 I only saw one. What are you seeing?
14:41 it's funny, i should probably jettison my desire to find deeper meaning in things like satisfaction
14:41 it's fine as a metric
14:41 Also, straightforward doesn't mean it measures what we want it too.
14:41 not everything needs to be an egg hunt
14:41 Statistically, I don't htink it is fine.
14:41 It doesn't have enough motion
14:41 it not sensitive enough
14:41 it takes a lot for people to knock a star off an app.
14:42 fast moving / short memory / in the now is better for id'ing trouble spots
14:42 "right now is awesome" yes / no + phone home wth current state
14:43 where we see that people say no if there is a hang.
14:43 or they are reading newspaper comments
14:43 I am open to the idea of multiple stats here... recommend (or satisfaction) as a long runner
14:43 and something short horizon as a 'oh shit' thing
14:45 so, true story... right now, my gmail is being a turd on reloading. I woud still give fx my normal satisfaction and recommend rating (maybe slightly
lowered)
14:45 but 'yo, this fucked' score would change
14:45 If you want eat dogfood on this, let me know, and let's up the addon to actually start asking some shit :)
14:46 maybe with a 'score' and 'why: [ ]' sort of thing
14:37 yes, i mentioned something like that
14:37 My hunch is that question I would want is something like "how are you: bad - ok - good"
14:37 or maybe 'how's it going'
14:37 heh
14:37 and let fx take the blame for everything
14:38 that way there is always some 'up' and 'everything' affects down
14:38 so in my experiment to quantify questions... you'd get two at a time.
14:39 so one can calibrate them directly
14:39 (two of the candidate questions)
14:40 leo's satisfaction questions are quite straightforward
14:40 I only saw one. What are you seeing?
14:41 it's funny, i should probably jettison my desire to find deeper meaning in things like satisfaction
14:41 it's fine as a metric
14:41 Also, straightforward doesn't mean it measures what we want it too.
14:41 not everything needs to be an egg hunt
14:41 Statistically, I don't htink it is fine.
14:41 It doesn't have enough motion
14:41 it not sensitive enough
14:41 it takes a lot for people to knock a star off an app.
14:42 fast moving / short memory / in the now is better for id'ing trouble spots
14:42 "right now is awesome" yes / no + phone home wth current state
14:43 where we see that people say no if there is a hang.
14:43 or they are reading newspaper comments
14:43 I am open to the idea of multiple stats here... recommend (or satisfaction) as a long runner
14:43 and something short horizon as a 'oh shit' thing
14:45 so, true story... right now, my gmail is being a turd on reloading. I woud still give fx my normal satisfaction and recommend rating (maybe slightly
lowered)
14:45 but 'yo, this fucked' score would change
14:45 If you want eat dogfood on this, let me know, and let's up the addon to actually start asking some shit :)
14:46 maybe with a 'score' and 'why: [ ]' sort of thing