hi, i'm yoni, and i work on a search engine.
i also look at customer complaints and suggestions

here's a thread on things iv'e learned from working with search engines and customers
1) people throw around the word "AI" like they know what it is outside of science fiction, when they don't

AI, or #machinelearning , isn't terminator robots, or the Dr. from star trek voyager (yet, IY"H it will get to be like the Dr.)
at the moment, it's more akin to a tree in
that, however you treat the seed, it will effect the tree. it cannot do more or less than the essence of what you put into it. which is why it can become racist if the programmers tint it to be so, and in regards to search, it learns English at the level of it's users
this is one of the reasons google is still the best at what they do. besides the hundreds of highly talented programmers working on their search engine, they have LOADS of years of data to use for making decisions, thereby leading to a better AI

2)computers understand data and
anything that is not data is an illusion made by humans to make us feel like a search engine understands us

i work with elastisearch or rather SOLR, basesd on the lucene library. if those words sound like techno-bable to you, that's okay.
it means that i work with a program that
uses other programs and a library of pre-written code to do all the hard work of finding stuff for me.
i then take this program that is doing the schlepping, and tell it where and how to look, and make it look pretty.
the first question you might be asking is "who wrote the
library that does the schlepping? maybe it's also biased?"

good question, assuming the library makes the final decisions, but it doesn't because it works with data. the search engine is a like a robotic arm controlled by a joy stick, it can do some cool stuff, but it can't
do whatever. it will only sort through the stuff you put in front of it
you give it cat pictures, it looks through cat pictures, you give it names of neonazis, it sorts through poop, etc.

3) search engines understand english.

related to above point, in short they don't. we
often think they do, because a lot of search stuff was written in english or in programming languages that use english words.

so, if you search for "THE THING NINJA USES TO GET MORE KILLS" and get stuff about Fortnite instead of Katanas, it's because current trends of searches
make the data about a guy who calls himself ninja who is getting kills in Fortnite, and not about japanese assassins and their tools

4)search engines are completely controlled by the owners

eh... yes and no.

while a search administrator theoretically CAN decide exactly what
the engine does and doesn't, quality search depends on good input of it's users. this means that a group of users who know how a search algorithm or protocol works, can manipulate it to display things the way they want.

this is what internet trolls did during Drumpf's running in
his first election. they hacked things, but not in the sense of bypassing security and changing the data. they simply shoved so much data at the search engines that the engine said "well, this is what most humans want, so i guess i'll give it to them"
could FB and google shut it
down? sure, but they didn't want to because that would make the engine too biased to the admin instead of the users. when they realized it was being manipulated in a bad way that undermined the intent, they took necessary measures

one thing i would like to add
not enough people know how to use a search engine properly, possibly because Google has gotten so good, that we don't question it.

if anyone want's i can give a short lesson on how, or you can check YouTube "how to power search on Google"

i believe if most people did,
there would be a lot less confusion and misinformation
You can follow @rabbiyo.
Tip: mention @twtextapp on a Twitter thread with the keyword “unroll” to get a link to it.

Latest Threads Unrolled:

By continuing to use the site, you are consenting to the use of cookies as explained in our Cookie Policy to improve your experience.