/sci/ - Science & Math » Thread #14187905

207KiB, 1056x1070, 1643570733594.jpg

View Same Google iqdb SauceNAO

Anonymous Sat 05 Feb 01:03:22 2022 No.14187905 View Reply Original Report

Quoted By: >>14187944 >>14188508 >>14192030 >>14192058 >>14192871

What quantitative methods and/or experiments would one use to prove that an online community is an ideological echo chamber reinforced by its moderators?

NOTE: this is NOT a /sci/ meta thread. This topic came up in discussion on another website and someone asked me to do an experiment. I need actual ideas for how to objectively measure the echo-chamberedness of an online community.

Anonymous

Anonymous Sat 05 Feb 2022 01:19:40 No.14187944 Report

Quoted By: >>14188010

>>14187905
>objectively measure the echo-chamberedness of an online community
not feasible even with easy-tier communities that have upboat mechanica, like reddit
sentiment analysis is too primitive to give reliable estimates of "closeness" of two posts

Anonymous

Anonymous Sat 05 Feb 2022 01:39:24 No.14188010 Report

Quoted By: >>14188066 >>14190046 >>14190469 >>14190474

>>14187944
Allow me to clarify the parameters of this experiment:

The hypothesis is that the website (a technology forum) has grown increasingly lopsided on the left-right political spectrum in recent years, at least partially as a result of moderation decisions. The site does have voting mechanics, although this is not reddit. It also has a "flag" mechanic that's somewhere between a downvote and a report.

I think maybe I can use a text classifier to tag posts as being "left", "right", or "neutral", and then tally vote and moderation statistics for each of the classes. I'm not sure how feasible that would be though.

Anonymous

Anonymous Sat 05 Feb 2022 02:03:04 No.14188066 Report

Quoted By:

>>14188010
>maybe I can use a text classifier to tag posts as being "left", "right", or "neutral"
again, too primitive to get anything reliable out of that
and i doubt that left/right can be used as an objective metric, since it isn't even well-defined
but you do you, it's a project for maybe a saturday night with tensorflow

Anonymous

Anonymous Sat 05 Feb 2022 04:41:55 No.14188508 Report

Quoted By: >>14189919

>>14187905
If you want to measure raw sentiment over time then that should be easy, if you want to determine WHY it's a circlejerk then that's more complicated.
Leftists don't care about your data though, they will engage in pilpul about how they won all the arguments or something. They have to be dealt with in the real world.

Anonymous

Anonymous Sat 05 Feb 2022 15:30:56 No.14189919 Report

Quoted By:

>>14188508
Agree, but every little bit helps. Having hard numbers to point to would be nice.

Anonymous

Anonymous Sat 05 Feb 2022 16:13:56 No.14190046 Report

Quoted By:

>>14188010
You fundamentally misunderstand how it works. People allign politically because they share beliefs with what they allign with. They don't pick a side and then copy the beliefs. You can't buy a political movement and change it how it suits you, and expect people change their beliefs accordingly. Instead, they will leave to who represents their beliefs better.

Anonymous

Anonymous Sat 05 Feb 2022 18:14:26 No.14190469 Report

Quoted By: >>14190474

>>14188010
Have users strongly self-identifying as either side rate things they perceive as belonging to the other side. Compare trends to see whether you can just take the average of those or if you need to adjust it (one of the "sides" can be overshooting temporarily for whatever reason).
Now have both rate the batch of things you believe to be the litmus strip and use the aforementioned bias correction.

Anonymous

Anonymous Sat 05 Feb 2022 18:15:54 No.14190474 Report

Quoted By:

>>14188010
>>14190469
And of course it cannot just be as simplistic as "my side", "their side", "not sure". The other anons have covered that but just in case.

Anonymous

Anonymous Sun 06 Feb 2022 00:26:00 No.14192001 Report

Quoted By:

bump

Anonymous

Anonymous Sun 06 Feb 2022 00:35:57 No.14192030 Report

Quoted By:

>>14187905
to put it abstractly, the formula is
2+2=5

Anonymous

Anonymous Sun 06 Feb 2022 00:44:57 No.14192058 Report

Quoted By: >>14192858

>>14187905
>prove that an online community is an ideological echo chamber reinforced by its moderators?
there are way too many parameters in your question. You an do any of the following separatedly :
- measure if a community is less ideologically broad than the average
- measure if the community is an echo chamber (and that concept itself is not very precise. Do you mean that it progressively radicalizes, or that it's ideologically stable over time?)
- measure if the moderation has an impact on the content (that would need a control group)

If you want to answer the full question of "is an online community is an ideological echo chamber reinforced by its moderators", you need a sociology study, and as interesting as the results may be, they probably won't be as manichean as what you're most likely looking for

Anonymous

Anonymous Sun 06 Feb 2022 05:45:08 No.14192858 Report

Quoted By:

>>14192058
>(that would need a control group)
Can't you just look at the moderation logs for each wrt each type of post?

Anonymous

Anonymous Sun 06 Feb 2022 05:53:42 No.14192871 Report

Quoted By:

>>14187905
first you need to make a few definitions. for instance, in your words, what is an echo chamber (and its relationship to communities that aren't echo chambers)

Capcode	All Only User Posts Only Moderator Posts Only Admin Posts Only Developer Posts
Show Posts	All Only With Images Only Without Images
Deleted Posts	All Only Deleted Posts Only Non-Deleted Posts
Ghost Posts	All Only Ghost Posts Only Non-Ghost Posts
Post Type	All Only Sticky Threads Only Opening Posts Only Reply Posts
Results	All Grouped By Threads
Order	Latest Posts First Oldest Posts First

Your latest searches