[e12e]

I'm not quite sure what's meant by "scholary papers", but I really enjoy(ed) Fielding's Ph.d thesis on "REST":

http://www.ics.uci.edu/~fielding/pubs/dissertation/top.htm

Mostly because, it's not actually about just REST, but the way he derives REST as a reasonable approach to architect hypermedia/hypertext "applications" (In quotes, because, he's not really talking about "web apps" -- he mentions some other patterns that do describe "web apps" though).

I have the impression few people read and understood his paper, and run around with REST like others run around with MVC. Which brings us to:

Trygve M. H. Reenskaug's "MVC" (neé Model-View-Controller-User): http://heim.ifi.uio.no/~trygver/themes/mvc/mvc-index.html

and, newer, less known: "DCI - A new Role Based Paradigm for specifying collaborating objects":

http://heim.ifi.uio.no/~trygver/themes/babyide/babyide-index...

I think that sums up the "papers" I generally refer back to, and find myself frustrated that so few people seem to have read and/or understood. Which leads to strange discussions and unhealthy re-inventions and "improvements".

Oh, I really enjoy some of the work of VPRI/Alan Kay -- but they've been rather thin on useful papers, as far as I can tell. I did enjoy a paper on Croquet's TeaTime protocol/world model -- but sadly I can't seem to find it... hang on, I think it might be this one here:

"Designing croquet's TeaTime: a real-time, temporal environment for active object cooperation":

http://dl.acm.org/citation.cfm?id=1094861

All less impressive than Shannon, Einstein, Knuth etc... but I really find those interesting.

[stiff]

The REST thesis, whatever merit it may otherwise have, is also a prime example of the academic style of writing going horribly wrong, though. You could communicate the important novel points of this text in 5-10 pages of clear prose, instead the thesis goes on and on and bathes in vague generalities, introduces lots of jargon that contributes little and so on. It reads like something from the philosophy department, or the kind of writing that Orwell mocks in "Politics and the English Language". This is the reason why to this day many people who use the term REST do not really understand what it means. Compare it to how Watson and Crick communicated their fundamental discoveries:

http://www.nature.com/nature/dna50/watsoncrick.pdf

[e12e]

I don't agree. If the point was to only document REST (and not the other architectures) -- then yes, it could have been a lot shorter. I think it gives a good overview of architectures in general, and ways to look at and analyse software. This way it makes a much better argument for REST -- and contrast it with other styles in common use at the time (and to a lesser extent, today).

I also don't think it is particularly wandering or obtuse.

> It reads like something from the philosophy department

Well, it is titled for "doctor of philosophy" ;-) (in Information and Computer Science).

BTW, how long is that DNA article? A cursory search only reveals articles that are cut off on the second page (at "inner-").

[deong]

Everyone I know with a PhD has a 10-page journal article they recommend you read instead of the thesis. For whatever reason, this is the way PhD theses tend to work these days.

[dmd]

My PhD thesis (U of Penn, cognitive neuroscience) is two of my 5 page journal articles, stapled together, with an introduction and conclusion added on mostly as an afterthought. I'm not sure why other schools don't do the same.

[deong]

I do get why some people dislike the sandwich thesis. From the standpoint of providing evidence that you deserve the PhD, it's fine. We judge research performance by published papers, so publish two or three good ones, print them out, and graduate.

If we're talking about the quality of the actual thesis as a separate document that has value on its own, the sandwich thesis isn't great though. No one would argue that it needs to be an entirely separate piece of work from your papers, but there's a legitimate case for requiring it to be pretty heavily edited into something resembling a book rather than a collection of articles on the same topic.