Thread by @anupampom

How Do you know you have made a blah/controversial tweet which is not in accordance with popular opinion.
A rough and useful metric is Reply/Retweet Ratio. If your tweet has higher replies than retweets, basically it moves against popular opinion/sucks.
esquire.com/news-politics/…

While Replies/Retweet Ratio is a rough metric . Its trend can be used to get an idea about ifluctuation in ideology/sentiment as well.

A Reply/Retweet Ratio>1 , generally indicates a blah tweet (meaning not appreciated by followers)

Expectedly Reply/Retweet Ratio is lower for Politicians.
An Exception @realDonaldTrump who through his tweets brought mojo back on twitter.
Lets plot his retweet/reply ratio over years.
(I Scraped his tweets and plotted Reply/RT ratio over years

We see while Trump indeed has a comparatively higher reply/RT ratio as compared to other politicians, His Ratio started increasing steadily post july 16, touched a high in first year of presidency. while has dipped a bit of late. A controversial politician indeed

Lets see the ratio for @narendramodi
A stark contrast from Trump. Generally below 0.4

What about ratio of @RahulGandhi
Rahul gandhi as well has a low Reply/RT ratio ever since he started tweeting(a bit late)

What is interesting here is that the ratio can be used to plot ideological/opprtunistics shifts indicators as well. Lets plot ratio of @rupasubramanya who has been accused by RW of making a U turn. You see a U on plot as well!(may 2014-june 2017). Ratio stayed <1

Ratio Plot of @dhume. Another U turn. Which started a bit earlier.

@tufailelif has also been accused of U turn by Indian RW. However there is no U in the plot. You see a distinct break after coronation of yogi adityanath as UP CM

Lets see the ratio of those who stayed the course.
@sagarikaghose consistently bad ratio , which worsened after her internet hindu coinage around 2011-12

Rana @RanaAyyub while not in same league as @sagarikaghose , Her ratio has remained above 1 consistently. Of late her ratio has dipped below 1, so it is possible people have unfollowed her/given up on her or she might have blocked the regular offenders

Let see Reply/RT plot of @rahulroushan , who started as the faking news guy and then crossed over with consistent political/RW stance.
His ratio has stayed constant<1. implying consistent messaging appreciated by followers

Technical Notes
1. Splines were used to smooth trends, at extremes(limited data,wider confidence bands) these cross<0 ,which is theoretically impossible , but overall GAM is a good representation of trends
2. A minimum of 20 k tweets randomly sampled from these account wa ss used

Indian Trump Kejriwal's @ArvindKejriwal plot looks like a veritable sine wave(though the ratio stays<1) cf modi/RG !..typically of his bipolar shifts