Loren Terveen

For years there's been a debate in my research community (social computing/HCI) about whether there is an ethical problem with researchers from tech companies publishing based on data only they have access to. There are a few important nuances to this debate: [Thread]https://twitter.com/oliviasolon/status/1341780460049985544 …

If we want to be all scientific to me it seems like the publication bias problem is much bigger than the replication/access to data problem? If there is *any* kind of prepublication review it's hard to defend as "science." And I think industry always has prepub review. 1/x

I was told that "scientists" working at tobacco companies can do whatever "research" they want but that it is not welcome in reputable journals. The prepub review problem is the issue -- only pro-tobacco findings come out of those labs, so publishing distorts the record. 2/x

Sometimes this is called a conflict of interest problem but calling it a prepub problem seems clearer. And if you cross out tobacco in my last tweet and instead put "social media" or "big tech" does that change the logic? Not sure but I don't think so. 3/x

I'm guessing Bell Labs / PARC work as industry-supported pseudo-"academic freedom" b/c the findings don't threaten the funder... or the funder is too confused/behind/absent to realize that they do. So there is then no censorship? i.e. no prepub review problem. End thought. 4/x

In the 90s at Bell Labs, we absolutely had to have our papers internally cleared before publication, although as I recall, it was for IP purposes rather than much worry about making the company look bad.

yeah this is what I've also heard from folks mostly now, which is why the Google thing was so surprising :-\

But Google never has had the kind of (more or less "pure") research culture that Bell Labs did.