api and readme

2 files changed, 59 insertions, 5 deletions

diff --git a/index.js b/index.js
index 9d46a85..68facd4 100644
--- a/index.js
+++ b/index.js
@@ -1,7 +1,7 @@
 var mastodon = require('mastodon');
 var pg = require('pg');
 
-var query = `SELECT id, favourites_count 
+var query = `SELECT id 
 FROM statuses 
 WHERE favourites_count > (
   SELECT avg(favourites_count) 
@@ -35,9 +35,7 @@ function cycle() {
         return console.error('error running query', err);
       }
 
-      result.rows.forEach(function(row) {
-        console.dir(row);
-      });
+      boost(result.rows);
 
       // disconnect the client
       client.end(function (err) {
@@ -47,4 +45,16 @@ function cycle() {
   });
 }
 
-cycle();
\ No newline at end of file
+var M = new mastodon({
+  access_token: process.env.AMBASSADOR_TOKEN
+});
+
+
+var boosted = {};
+function boost(rows) {
+  console.log('Boosting the following:');
+  console.dir(rows);
+}
+
+cycle();
+setInterval(cycle, 1000 * 60 * 15);
\ No newline at end of file
diff --git a/readme.md b/readme.md
new file mode 100644
index 0000000..98d7b19
--- /dev/null
+++ b/readme.md
@@ -0,0 +1,44 @@
+# Ambassador Bot
+This is my attempt to fix the problem of followbots on Mastodon, while also adding some potentially useful and interesting behavior to a bot that can be run on any server.
+
+The AmbassadorBot will live on your server and find those local toots that have earned a high number of favs relative to other local toots. It will then boost them.
+
+The idea is that it's always boosting the 'best' tweets of the instance that it run on. Anyone who follows your instance's Ambassador will therefore get the best toots of your interest in their federated feed. No need to use followbots to systematically follow users when you can get the best of their tweets easily!
+
+## Installation
+First, you'll need to create a new account on your instance and use [the @tinysubversions extractor](http://tinysubversions.com/notes/mastodon-bot/) to get an OAuth token for it. 
+
+This bot has to be installed on your instance server, so unless you're the admin you're not going to be able to set it up yourself. The reason for that is that the bot reads directly from your database, rather than using the API. It requires the following environment variables (and uses the provided defaults when they're missing):
+  
+  DB_HOST (defaults to '/var/run/postgresql')
+  DB_NAME (defaults to 'mastodon\_production')
+  DB_USER (defaults to 'mastodon')
+  DB_PASSWORD (defaults to '')
+  AMBASSADOR_TOKEN (no default)
+
+To install it, set your environment variables and do the following:
+
+  git clone git@github.com:mbilokonsky/ambassador
+  cd ambassador
+  yarn
+  yarn start
+
+It'll run every 15 minutes, boosting new toots that have crossed the threshold.
+
+## How does it determine what's good enough to boost?
+So, this is still sort of an open question but right now I'm using the following query:
+
+```SELECT id 
+FROM statuses 
+WHERE favourites_count > (
+  SELECT avg(favourites_count) 
+  FROM statuses 
+  WHERE favourites_count > 1
+  AND created_at > NOW() - INTERVAL '30 days'
+)
+AND created_at > NOW() - INTERVAL '30 days';```
+
+It takes an average of all toots with 2 or more favs over the past 30 days. Any toot within that window that has more than that number of favs gets a boost. Note that most toots won't get 2 favs - so this is already filtering out most toots in your instance. The hope is that by averaging what's left and picking the top half we'll end up with a pretty high standard for what gets boosted, but this algorithm will be tweaked over time.
+
+## Seriously? You want me to give this thing access to my production database?
+Look, I get it - but how else do you want me to find your top tweets in a performant way? I'm not passing any user input into the database, just repeating a static query. I am not, btw, a database expert - I pieced this query together through trial-and-error and if you want to propose an optimization I am all ears.
\ No newline at end of file