3 ms·
This makes me think of the (apparently true?) story of the shuffle mode in iTunes being seemingly too random for people to the point that some believed it was b
by quibono 5y ago
This makes me think of the (apparently true?) story of the shuffle mode in iTunes being seemingly too random for people to the point that some believed it was broken. I have a feeling the differences between the theoretically optimal path (i.e. the one from the start of the article) and the improved ones they get by including additional heuristics are there mainly because they make the drivers feel like they're working more efficiently. Which is very fair considering that the way the drivers feel about the routing is also a cost.
- leetrout 5y agoAt a previous startup I worked at I suggested letting users paint weights on a map to influence routing and clustering. Nothing specific, not anchors like additional stops on google maps but like "this is a neighborhood to locals" that would influence clustering and "avoid this area" for routing so the algorithm would use interstates or not based on what appeared more optimal even with just using time or distance as a cost.
- quibono 5y agoI think this is the way to go whenever optimization software meets real humans. I've had some experience with trying to optimize diets based on nutrition, and most of the work goes into making sure the suggested improvements aren't too radical and that the new 'improved' diet feels close to the original. Simply because the most optimal diet is something that no one would actually want to eat, while dramatic changes in suggested foods will simply be ignored by most.