Two items posted on Linked in during July 2017 about algorithms not performing as expected.
Some thoughts on “algorithmic dysfunctions” (posted July 2017)
Algorithmic dysfunctions? Annoyance turning to concern in how algorithms are handling searches. Previously noted how searching LinkedIn for a name (given+family) yields results without part or all of the name. Puzzling at best – isn’t LinkedIn first and foremost a social network of people with names?
Also noting continued degradation in Google searches – even using “verbatim” and quotes around terms, get results without those exact terms.
Algorithms interpreting what one asks for based on some statistical model? Systems getting so fancy they eschew the simple and direct route in favor of some circuitous tour by default?
I do appreciate that the amount of data is vast and increasing, and that it’s pretty amazing what we can do with searches of it. However, for my purposes at least, it really seems like optimal search engine performance was reached a few years ago and has been declining since. Where is this going?
Cathy O’Neill asks important questions re algorithms in our future & proposes a 4-layer hierarchy of bad algorithms.