3 ms·
it confuses what i liked in the past with what i might enjoy in the future. or what might feel familiar or what someone like me liked. how am i ever supposed
by basch 9y ago
it confuses what i liked in the past with what i might enjoy in the future. or what might feel familiar or what someone like me liked.
how am i ever supposed to have new foreign experiences if recommendations are based on past experiences or grouping me into demographics to make me more like others?
sure everyones a mimicry or pastiche of others, but its losing its randomness and becoming more uniform and standardized.
- Houshalter 9y agoA perfect recommendation algorithm would be optimal in the exploration/exploitation tradeoff. Suggesting new things with some probability to see if you might like them or not. Most recommendation algorithms today are way too far on the exploitation side of the tradeoff, but it's not a requirement.