Have you ever visited tvtropes.org? Don't go if you don't have the rest of the day to waste. It is seriously addictive, mostly because it's a sterling example of hypertext in action, maybe even better than Wikipedia. Every article is linked to the others. Every citation has a link. You want to check a reference, you click. You want to surf a chain of references N generations back, you click N times. You can't not follow the links on tvtropes.org, unless you simply hate all pop culture.
It is the height of irony that HTTP was originally invented to link up academic publications, because when last I surfed academic journals, six years or so ago, it didn't work that way. You would click a journal article, maybe click through a couple of crappy login screens, maybe navigate up and down some menus on some journal's CMS, and finally find the article after a bunch of surfing. It would have footnotes. The footnotes were almost never links. Instead you had to manually type their key information into your local library's website in some other window, and then hope that your local library had the journal, then surf through some more screens and up and down some more delightful CMS cruft before you found the article.
Smart folks would download PDFs of everything as they went and then file them in some personal database like Endnote. But they still wouldn't be cross-linked. You had to spend time organizing them, or use the search a lot.
I assume that this workflow has improved at least a bit since then. Maybe Endnote or its competitor is smarter, maybe journals have finally started publishing hyperlinks. Maybe it has even improved to the point that academic literature is half as easy to surf as, say, tvtropes.org. In which case I take back much of what I said, and we can go back to estimating how many people can actually be expected to access the blessed terminal where all of this can happen. But I bet it hasn't improved that much. I bet it's still impossible to construct a hyperlink from Nature directly to Science, mostly because they've got different and incompatible DRM.
Mind you, things are heaven compared to when I was in grad school, when following references from a paper meant going into the stacks, laboriously collecting one heavy paper volume per footnote, physically hauling them back to a photocopier, and copying the articles two pages at a time. Oh, the hours I spent. Life is much better now. But, you know, young people don't know or care about how hard I had it. All they can see is that it's fifty times easier to research the plot of any episode of Neon Genesis Evangelion than it is to follow a chain of citations from a Nobel-winning publication back two generations, and that a popular article about a videogame from 1980 can link to its sources but a popular article about a publication from last Wednesday's Nature can only link to a paywall.
Science is always going to be a hard sell compared to the rest of our culture, so why do we make it even harder?
It is. try google scholar, pub med, or academic.research.microsoft.com for starters and Mendeley for building a personal "library" (note Mendeley isn't perfect but it is a lot better than not using a paper manager.) Things aren't great but they are better.
> I bet it's still impossible to construct a hyperlink from Nature directly to Science, mostly because they've got different and incompatible DRM.
I have no idea I don't read Nature or Science. The Computer Science literature I read is all easily searchable these days.
> Mind you, things are heaven compared to when I was in grad school, when following references from a paper meant going into the stacks, laboriously collecting one heavy paper volume per footnote, physically hauling them back to a photocopier, and copying the articles two pages at a time.
I still do this (but for Religious Studies research in my spare time) and don't really mind it, although I try to avoid photocopying when possible as it is a pain.
> All they can see is that it's fifty times easier to research the plot of any episode of Neon Genesis Evangelion than it is to follow a chain of citations from a Nobel-winning publication back two generations, and that a popular article about a videogame from 1980 can link to its sources but a popular article about a publication from last Wednesday's Nature can only link to a paywall.
I am usually against the idea of progress, but in this instance I do believe this will not be the case forever. Already we are starting to see changes in the publishing industry. Search tools are improving, archiving tool are improving, library collection tools are improving. Yes it is slow, but the problems they are solving a much harder than the problem of generating new content. New content can be put the days leading format (papyrus, scrolls, codex, books, pamphlets/newspaper/magazine, websites/blogs/wikis) at zero extra cost. Old content however costs a lot to update. Libraries and academic journals have a lot of content that existed before the web, before the pdf, and before the wiki. It takes sometime of the format transition but if the past is any guide the old formats eventually become historical curiosities rather than knowledge repositories.
Unfortunately, I was never a CS guy, I was a cancer researcher. And if you can't read the Nature journals you're not going to keep up with medical or genetic research -- everyone in the field openly lusts after Nature publications; Nature gets first crack at every manuscript -- and Nature sits firmly behind a paywall.
Let's try your helpful suggestions on the legendary paper, Viable offspring derived from fetal and mammalian cells, a.k.a. "the Dolly-the-sheep paper", Nature 385, 810-813 (27 February 1997). Surely I can write a blog post in which I link directly to the methods section of this famous work, which contains the first-ever working recipe for cloning an entire mammal? I mean, this is from 1997. We even had the web in 1997. Don't try to tell me that this has to be painstakingly transcribed from some faded papyrus.
Pubmed has the abstract, of course -- they have abstracts for everything:
And Microsoft provides lots of cute metadata about this paper, but no actual contents:
...so if I want to talk about the methods section, it looks like I'll have to tell my blog readers to click through to, e.g., the Nature Publishing Group and pay thirty-two dollars for the privilege :
Or will I? Google Scholar to the rescue:
The first link is to a reprint of most of the article, in a book. And they've managed to include every page. Today, anyway. For me, this time. (Other pages of the book are "not included" in the preview, so maybe others won't be so lucky?)
Oops, except the figures and tables aren't there. Sure hope there wasn't any data in those figures and tables.
So I'm saved, sort of, thanks to Google's big scanning budget and even bigger legal budget, provided I want to cite the words of Wilmut et al, and not their data. Or their slightly-more-obscure references, or any of their many citations -- for those, I will have to repeat this dance, and hope I still get lucky.
I know I've got a bit of an idee fixe here, but I cannot get over how utterly lame this is. My god, it is lame. Much of the best work of modern civilization, the pinnacle of hundreds of years of scientific advance, work which our society invests billions of dollars in, is hidden inside this pitifully obsolete rent-seeking system while we use state-of-the-art infotech to exchange pop-music lyrics and annotate famous games of Magic: The Gathering.
 No, that is not thirty-two dollars per year, or even per month. That's one paper. You're probably better off springing for a subscription ($199) although that only gets you Nature: all the other journals in the Nature family, like Nature Medicine, are extra, and of course every other journal is also extra.
And the most infuriating thing: I'm already paying for most of the public research with my taxes.