On one side, there are some who say the future of scholarly communication in science is databases – or, rather, more or less shared and curated data sets. Some of the folks in this crowd go farther to say that science is a continuous stream and people should be able to comment on and point to this stream.
There are those who see the disaggregation of the journal with the papers remaining more or less the same. So databases of discrete pieces that can then be re-aggregated (I’ve mentioned this before)
And there are those who basically think we’ll sort of go on as we have been, but perhaps with more interconnectivity, supplemental data, more multimedia, and so forth. (see, for example, discussions surrounding the “article of the future”).
Cameron Neylon wrote a post on this in response to a discussion at Science Online London ’09. I basically agree and I said something similar in an earlier friendfeed stream. Journal articles tell a story. They place the work in a context and within a body of work and they are complete in describing one bit of knowledge. IMO there will continue to be a need to stop, reflect, and tell a story. To have some results and to discuss them, even if they are part of a larger project. I agree, as some others have been saying, that if we can build trusted repositories at the data (at the research institution), then it will become less important that the supplemental data be included in the journal. (I suppose in some areas you put your data in a disciplinary repository, not in supplemental data already).
Sure, we can see our life’s work as a stream, but in practical terms, we need to modularize and stop and start sometimes. How can we learn from others’ work if they only make a data stream open, without discussion or with continuous discussion. This isn’t even going into the whole Kuhnian revolution thing.
I also think that we’ll have to have some sort of certification/filter apparatus. If not journals, then something else. It isn’t scalable to say everyone should read all the articles that come out and then later we can use downloads or voting as a measure of utility, value, or quality. Everyone can’t read everything – they might be able to keep up with browsing the titles of one area of something like ArXiv, but I suspect from the several articles correlating placement in the alert with citedness/usage, that people get tired before getting to the end of even that. Relying on your friends and colleagues isn’t good enough, either. They certainly have limited time and energy and also their own research goals which are not the same as yours. Nor is full text natural language search/alerting the (full/complete) answer. All of these would stifle some interdisciplinarity and creativity. The serendipitous moment when you find something really good in someplace unexpected. Perfect precision is not the goal!
Journals grow around a community and a community’s interests. The editors guide the development of the journal to reflect the needs of that community. Sure, they are somewhat conservative. When a part of the community isn’t well served, they start a new journal.
Hm. well, nothing gained in this post! No doubt I’ll keep returning to this.