Promotional emails. Often when I purchase something from a new website, I will start getting two or more promotional emails a day from that company. It is excessively annoying and often leads me to open the third or fourth email looking for the unsubscribe button, which is not the engagement I think they want, but it is what the algorithm seems to be measuring. Also, they might be reading every preview open from email apps as an actual open. My use of Evite and PaperlessPost suggest that might be the case. So the companies think they are getting positive engagement from these bombardment campaigns while they are actually alienating new customers.
... and hard-copy charity appeals. My spouse swears that a bunch of charities we've donated to have spent at least the amount of our donation in more fourth-class mail asking for more money. If we haven't donated with a handwritten check or other hardcopy evidence, why do we keep receiving them? (At least it helps fund the USPS, but still...)
I think we also get used to data modeling fails in consumer services and work around them. For example, I know rideshare apps are going to route cars to the wrong side of my street, so I'll request to a different spot nearby. It's a pain, but there seems to be no alternative - "improve your routing" is rarely a customer feedback option, and if they haven't done it in 10 years they are not likely to start now.
A question on how much you're interested in the modeling and how much in the measurement. In traffic engineering, I'm pretty sure all traffic flow measurements are by vehicle, not by person trying to get somewhere -- that's easier to measure, but it obviously overweights single-occupancy vehicles. And it completely warps how transit is designed in North America.
Vocational surveys putatively helpful to guiding people into careers - not the ASVAB (I'd love know the internal research on that) but other instruments that career counselors use. For a moment we can ignore the sociological literature on how it feeds the reproduction of inequality to tell someone what they're destined for, but I see the process of taking it and then receiving guidance as a form of (very crude) modeling.
Application Tracking Systems for hiring - companies are flooded with spam, and then the process for candidates devolves into the meta-game of "beating the filter" and shotgunning hundreds of applications. Waste of time and dehumanizing for all involved.
Promotional emails. Often when I purchase something from a new website, I will start getting two or more promotional emails a day from that company. It is excessively annoying and often leads me to open the third or fourth email looking for the unsubscribe button, which is not the engagement I think they want, but it is what the algorithm seems to be measuring. Also, they might be reading every preview open from email apps as an actual open. My use of Evite and PaperlessPost suggest that might be the case. So the companies think they are getting positive engagement from these bombardment campaigns while they are actually alienating new customers.
... and hard-copy charity appeals. My spouse swears that a bunch of charities we've donated to have spent at least the amount of our donation in more fourth-class mail asking for more money. If we haven't donated with a handwritten check or other hardcopy evidence, why do we keep receiving them? (At least it helps fund the USPS, but still...)
Recipe times are always insanely optimistic. I don't know anyone who can actually match them outside of professional chefs.
I think we also get used to data modeling fails in consumer services and work around them. For example, I know rideshare apps are going to route cars to the wrong side of my street, so I'll request to a different spot nearby. It's a pain, but there seems to be no alternative - "improve your routing" is rarely a customer feedback option, and if they haven't done it in 10 years they are not likely to start now.
I forgot the Apple Watch and falls! In martial arts class, pick Martial Arts as workout option and the watch still asks if I’m okay when I’m thrown.
Academic library "one search" box results.
Netflix (and all streaming service) recommendations
Child development charts that pediatricians rely on!
A question on how much you're interested in the modeling and how much in the measurement. In traffic engineering, I'm pretty sure all traffic flow measurements are by vehicle, not by person trying to get somewhere -- that's easier to measure, but it obviously overweights single-occupancy vehicles. And it completely warps how transit is designed in North America.
Vocational surveys putatively helpful to guiding people into careers - not the ASVAB (I'd love know the internal research on that) but other instruments that career counselors use. For a moment we can ignore the sociological literature on how it feeds the reproduction of inequality to tell someone what they're destined for, but I see the process of taking it and then receiving guidance as a form of (very crude) modeling.
Application Tracking Systems for hiring - companies are flooded with spam, and then the process for candidates devolves into the meta-game of "beating the filter" and shotgunning hundreds of applications. Waste of time and dehumanizing for all involved.