The quality of science textbooks has been decried by researchers for decades. Even with SER focused on narrowly incremental improvements, on the easily deployable, textbooks often don't reflect it. Incentives are poorly aligned, authorial resources too limited, and scope constraints too severe. Further, attempts at less incremental, more foundational improvements, are very rare (eg, integrated life-sciences curriculums). What support might allow OER to achieve better?
OpenStax "View online"'s nice text search allowed quickly finding such common misconceptions. And OpenStax supports open errata databases. The database of a past version of a textbook could be reached by URL editing. One misconception was pointed out twice, and WONTFIXed "Our reviewers determined the textbook meets scope, sequence, and accuracy requirements as is. No change will be made." Perhaps if the errata database had a discussion mechanism, like github issues, the problem might have been clarified and the new version of the textbook fixed? (Disclaimer: It's a misconception I commonly check for, and one of the two comments might be mine.)
Problems this pervasive are structural. How might OER creation be restructured to work better? Ideally much much better? For instance, support for easily combining patch-sets might allow an ecology of "ok, I fixed/improved this bit", avoiding multi-year gatekeeping delays.
A decade or two back, NSF almost funded a national science education wikipedia... but didn't. Each common misconception might have had pages of literature overview, discussion, remedies. Is there an alternate timeline where science education content has dramatically improved?