SK: why [do] you think it's significantly better than the current replication+extension approach?DR: I can see the arguments on both sides here. I definitely support replications and sometimes it may make sense for the author to “start a new paper” rather than make this an improvement if the old one. I also think that the project should be time stamped, evaluated and archived at particular stages of its development.But I lean to thinking that in many to most cases a single project with multiple updates will make the literature clearer and easier to navigate than the current proliferation of “multiple very similar papers by the same author in different journals”. It also seems a better use of researcher time, rather than having to constantly restate and repackage the same things
PS (?): Are these are things like 'living' google docs that keep getting updated? If so I'd consider using workarounds to replicate their benefits on the forum for a test run (e.g., people add a version to paper title or content or post a new version for each major revision). More generally, I'd prefer the next publication norm for papers to be about making new 'versions' of prior publications (e.g, a 'living review' paper on x is published and reviewed each year) than creating live documents (e.g., a dynamic review on x is published on a website and repeatedly reviewed at frequent and uncertain intervals when the authors add to it). I see huge value in living documents. However, I feel that they wouldn't be as efficient/easy to supervise/review as 'paper versions'.They might also not be ideal for citing as they would be an ever changing resource. I can imagine the whole academic system struggling to understand and adapt to such a radical innovation given how focused it is on static documents. With all of this considered, I'd like 'dynamic/living work' to be incentivised with funding and managed with informal feedback and comments rather than being formally reviewed (at least for now). I'd see living review work as sitting alongside and informing 'reviewed' papers rather than supplanting them. As an example, you might have a website that provides a 'living document' for lay people about how to promote charity effectively and then publish annual papers to summarise the state of the art for an academic/sophisticated audience.
I don’t think living documents need to pose a problem as long as they are discretely versioned and each version is accessible. Some academic fields are/were focused on books more than papers, and these were versioned by edition. Preprinting is also a form of versioning and combining the citations between the published paper and its preprint/s seems to be gaining acceptance (well, google scholar may force this by letting you combine them) - I don’t recall ever seeing a preprint citation indication a specific version (on preprint servers that support this) but its seems possible.I think the issue with the current citing practice for live documents like webpages is that even if a ‘version’ is indicated (e.g. access date) past versions aren’t often very accessible.
I mainly agree with @gavintaylor, but I appreciate that 'changing everything at the same time' is not always the best strategy.The main idea is that each version is given a specific time stamp, and that is the object that is reviewed and cited. This is more or less already the case when we cite working papers/drafts/mimeos/preprints.Gavin, on the latter 'past version accesibility' issue, This could/should be a part of what we ensure with specific rules and tech support, perhaps.