Sheri

its worth fighting for 🌷

Writer of word both truth and tale. Video producer, editor, artist, still human. Hire me?

Check #writeup for The Good Posts.
β€”

Slowly making a visual novel called We Will Not See Heaven, demo is free. Sometimes I stream, or post adult things. Boys' love novel enthusiast. Take care, yeah?

πŸ’ŸπŸ’ŸπŸ’Ÿ
TECH CAN ONLY BE AS KIND TO US AS WE ARE TO ONE ANOTHER.


πŸ–₯️ blog
sherishaw.net/blog

Sheri
@Sheri

any fiction writer, character artist, designer. any researcher or journalist, and all the kindest friends who find the question's answer before you even have time to search yourself

we've all googled something one fucking time and had your targeted ads think you a person entirely new

i have researched hair types that are not my own, i have looked up image references of vitiligo, i buy femme and masc clothes. i research countries i do not live in and businesses i would never patronize.

yet the interface i use to search is a business funded largely by selling ads and associated data, however indirectly

a search engine's priorities laying in profit is bad, we all know this. but it's also insidious thinking, this assumption that knowledge only exists to build a data profile on a user.

i live in objection to the capitalist assumption that one would only ever seek knowledge self-applicable. that first and foremost, the reason to look up information is to, yourself, capitalize on it.

i don't want ads to represent a false me, a version built from simply wanting to understand.

i don't want ads to represent me at all.

...and if i see one more goddamn Hims/Hers ad because google can't decide if i'm a guy or a girl, i'm going to Secret Third Thing


ireneista
@ireneista

for those who don't know, we spent five years on Google's advertising privacy team and we have a thorough understanding of online advertising infrastructure, which is a complex topic that even engineers who haven't worked in that particular field usually underestimate.

on our own background knowledge, we agree that the assumptions described above are baked in to these systems.


You must log in to comment.

in reply to @Sheri's post:

in reply to @ireneista's post:

this... means a LOT to read thank you? especially from y'all!
i really hope you all made it out of google's "fun work atmosphere" sane, i don't think i could stand to work for a tech company let alone a 'cool' one

but i'd sooner blame the 'company' part then 'tech', i guess?

as a mental health activist, "sane" is the one compliment we can't accept, but thank you :) we weren't sane going in and we aren't sane now

but yes we made it out. our departure was pretty epic, and entwined with the fact that the company now has a union.

and you're very welcome! you did a good job of speaking to the underlying incentive structure and practices, rather than to technical details that weren't available and would have been distractions if they were.

that's a good point! i should work that out of my lexicon when i guess i mean... 'empathetic?' out on the other side without giving up on the world

i'm terrible at math! which is why i write the feelings and the patterns of behavior, and cite the experts, comme vous

I'm tempted to go a step further and add that there are deep statistical/computational limitations that slays the dream of "advertisomancy" - take any collection of covarying predictors and when introduce the possibility that they vary over time, you've not only made your job harder, you've probably made it impossible unless you can collect much, much more data and never, ever stop doing so. Even if the measurements are good (which they generally aren't), determining the temporal relevance of any given datum ("this data is relevant for the next two days, that data will till be relevant in two years") is a supernaturally difficult problem.

Which is to say, it's not merely that these limiting assumptions are baked in - they're kind of the oven itself.