Introducing Echowings — Diversify your Political Twitter
For a large chunk of US citizens, the result of the 2016 Election was nothing if not surprising; after all — a wide variety of news outlets had Hillary Clinton pegged to win with overwhelming odds.
Following the Election, those same news outlets looked inward, and worked to understand how data had failed their prediction.
A lot of the blame was given to the “social media echo chamber” — that emerges when a social media user tends to follow other accounts & be served ads on the platform that have a similar social outlook to their own.
Like any good tech company, we started to think how we could help empower folks to fix this problem!
Break out of your echo chamber
So, ahead of the inauguration — we’re introducing Echowings:
It’s a simple site that uses very basic Machine Learning & Natural Language Processing to process the underlying sentiment of over 750,000 public tweets, published in the first 24 hours following the election.
In plain speak — Echowings loosely approximates which active Twitter users were happy & sad about the results of the 2016 Election.
Echowings uses this loose interpretation to send you recommendations of Twitter users to follow with an opposite political leaning to your own.
For the Nerds
The code for Echowings is available on Github:
By all means — fork it, hack it, bend it. It is written in a way that allows any developer to modify the search term, and have it relate to any topic on Twitter.
If you find bugs, or places for improvement and accuracy — we would encourage you to please contribute back to the project!
Disclaimer
Of course — Echowings is totally experimental, not guaranteed to be accurate 100% of the time. For that reason, the results served by Echowings should absolutely not be taken as definitive.
The basic nature of the underlying technology (and it’s inability to detect subtle language quirks, like sarcasm) means that we’re occasionally wrong. We haven’t studied this at length, but we’d estimate Echowings to be correct around 75% — 80% of the time.
Echowings only uses public information that is freely available through the Twitter API — however if you find Echowings has categorized you incorrectly, please notify us at hello@echowings.org — we’d be happy to either recategorized you, or completely purge your record for the system.
Thanks! Check it out and let us know what you think.
Love Hugh & the Sanctuary Computer Team
