Sociable Media
I've had an on/off battle with twitter for the past few years.
On the one hand it has been an invaluable service for me: introducing me to some wonderful people, finding me work and pulling me up when I've been down. But on the other hand it is often a toxic echo chamber filled with negativity.
I also find that the website itself often exacerbates an already negative mental health with its demand for continual engagement, like a needy toddler yearning for the dopamine nipple. Terrible metaphors aside I was wondering about this tension - between an ecosystem who's community I love but an environment I hate - and possible ways round this.
I set out some basic ideas for a quick prototype:
- a slower web: rather than constant reloading of media, the page should refresh at the users request and only irregularly,
- original content: stripping out noise to only show original material from those people I am following,
- positivity: use filtering to provide a more positive experience through showing only tweets matching certain critera
Using the twitter API I had a pretty good basis for fetching and showing data based on my user timeline and adjusting the response based on these three criteria.
By stripping out asynchronous requests we can slow down our experience and put a stop to the bottomless and compulsive experience of chasing the infinite scroll. We can then also exclude any replies and filter out any retweets to focus on original content.
In terms of focusing on more positive content I've played with interpreting twitter data before in using the AFINN method for analysing sentiment in tweets about Christmas. Basically a text string is broken down into words which are then compared to a predefined dataset that scores words based on the emotional response they reflect. For example, 'bastard' has a score of -5 whilst 'outstanding' has a score of 5.
With this criteria/method we can gauge whether a tweet is overall positive or negative and assign it a score and exclude those that fall below a certain benchmark (eg 0 which is - essentially - neutral emotionally).
So I had a play over a couple of evenings and this is what I came up with: the AFINN twitter experiment.
The sentiment filter is pretty manageable but sadly doesn't work with sarcasm so there are plenty of false positives (and false negatives too). Also am fully aware that this experiment removes one of the core features of twitter: community. I've provided jumping off points for engaging with individual tweets but as a proof-of-concept I think it goes someway to achieving what I set out to do.
Trigger warning - this plaything shows tweets from my own timeline (with tweets from protected/private accounts removed) so primarily shows content about tech/comics/politics.