Thread by @neuralreckoning

One of the things we're most excited about for #nmc3 is our new approach to reducing bias by eliminating editorial selection. We're replacing it with a scheduling algorithm that ensures people get to see the science they're interested in. Want to hear more? (1/10)

https://twitter.com/neuromatch/status/1308767994861019136

One of the big sources of bias is editorial selection. Conference organisers select a subset of talks to be featured in single tracks, and others are given posters or multi-track sessions. We wanted to find a way to eliminate this bias. (2/10)

At #nmc3 every submission gets a talk. After submission closes, we'll ask participants to go through a (blinded) list of abstracts to select the ones they're interested in. Using participants' and presenters' free times, we'll automatically build an optimal schedule. (3/10)

Some talks will be of wide interest, and they'll end up being scheduled close to single track, and others will be more specialised and they'll end up multi-track, but at times that don't conflict with other talks that are interesting to the same audience. (4/10)

We'll also group together related talks using the same topic modelling algorithms used for neuromatching in previous iterations of NMC (and CCN). In particular, the "interactive talks" (5m+10m Q&A) will be grouped into 2h sessions to make communities of talks (5/10)

The scheduling problem can be posed as a massive integer programming problem with potentially millions of decision variables and constraints, and a very complex objective function taking into account the number of watched hours, and thematic grouping. But it can be solved. (6/10)

Another problem is gathering people's preferences. We can't ask everyone to read all the abstracts submitted, so we will have an abstract browser with manual keyword search and automated suggestions based on topic modelling. This will be quick and painless. (7/10)

A side benefit of all this is that for everyone who gives their preferences, we will generate a personalised suggested schedule including alternatives for each hour of the conference. (8/10)

We're hoping it will lead to one of the most diverse, inclusive and interesting neuroscience conferences ever, where the aim is not to have the flashiest talks highlighted, but for everyone to be able to see the science that they are most interested in. (9/10)

Thanks also to the amazing efforts of @titipat_a, @tulakann and @patrickmineault for heroic efforts putting everything together in no time at all to make this possible. See you at #nmc3! Don't forget to submit your abstract by Oct 2. (10/10)