,
10 tweets,
4 min read
Read on Twitter
Thread! Thanks @reyammer for and @matthew_d_green for this eye-opening tweet: . @reyammer's thread is about [1/10]
double blind vs. pre-acceptance pre-prints/disclosures and problems around this. Let's discuss a second much bigger problem, a fundamental problem introduced by our review system that I never really thought about until @matthew_d_green pointed it out. [2/10]
This is not about single- vs. double-blind, single-blind wouldn't solve this. It is basically the second half to the problems pointed out by @socrates1024 here: twitter.com/socrates1024/1…. But enough disclaimer, let's go: [3/10]
In our field it's not uncommon that groups work on a largely overlapping idea ultimately one group scooping the other. Research takes time, so if both groups start at the same time it could be difficult to predict who wins the race (both groups not knowing of each other). [4/10]
Here our review system gives PC members (+teams) a highly unfair advantage over everyone else. Knowing what is not yet public but in submission is highly valuable. It allows you to avoid starting work on a colliding idea when the idea comes up, stop early, or change focus. [5/10]
Papers are in submission before public for around 2 months and multiple years depending on venue and how many resubmissions until accepted. Worst case: you invest many months or years in a project before discovering it was already done by someone else just not public yet. [6/10]
You probably know this from your reading group: you read a paper together and multiple people have the same great follow up idea. This brings us to the second problem and this is something everyone should be absolutely aware of: [7/10]
If you read a paper that just became public and have your idea moment, all the PC members (+groups potentially) that read the paper not unlikely had that moment months or years ago when they read the paper. Again, newcomers and small groups have a huge disadvantage. [8/10]
Third problem: PC members see if submitted non-public papers overlap with work (in progress) from their group. This allows them to do damage control with time-stamping or submit to a smaller venue to get the paper even through peer review before the other is over. [9/10]
There are more advantages, but, what is the solution? Should we mandate "ePrint/arXiv before conference-submit" to eliminate these advantages or prohibit "ePrint/arXiv before conference-accept" to reinforce these advantages for PC members as gratitude for their service? [10/10]
