Duplicate entries found in output

[JD-P]

Right now the outputted predictions have a lot of repeats, for example:

Researchers discover a black hole feeding frenzy that breaks records: 15.187169188969229%
http://sciencebulletin.org/archives/10085.html

Google Test of AI's Killer Instinct Shows We Should Be Very Careful: 17.193493474567934%
http://gizmodo.com/google-test-of-ais-killer-instinct-shows-we-should-be-v-1792256809

My aperiodic rhombic bathroom: 15.745731500468352%
https://liorpachter.wordpress.com/2017/02/11/my-aperiodic-rhombic-bathroom/

Twitter only grew by two million users during Trump mania: 15.384931115924893%
http://www.recode.net/2017/2/9/14558890/trump-twitter-user-growth

Researchers discover a black hole feeding frenzy that breaks records: 15.187169188969229%
http://sciencebulletin.org/archives/10085.html

Google Test of AI's Killer Instinct Shows We Should Be Very Careful: 17.193493474567934%
http://gizmodo.com/google-test-of-ais-killer-instinct-shows-we-should-be-v-1792256809

---

My current hypothesis is that this is caused by upvotes being clustered in the same time period, so the method of grabbing stories near in time to an upvote means you end up grabbing the same stories multiple times.