I read <i>Future Shock</i> by Alvin Toffler for the first time a couple of years ago, on its 50th anniversary. It's an interesting read on a number of points, both in documenting the history of information overload, but also in laying out predictions for what the future might hold. As with most sets of predictions, some are more accurate than other, though the patterns there are themselves interesting.<p>Ultimately, time is finite, attention is rivalrous, and information is not infinitely actionable. It also varies tremendously in quality, and different filter and incentive systems modulate that quality strongly. (This last fact itself is part of the pattern of hits and misses in Toffler's list of predictions.)<p>I've started to note that the DOI is the new URL. With all the issues of academic publishing, there's still generally a higher signal:noise ratio in that than in general Web content. Likewise books and other forms of traditional, gate-kept publishing. Much of the better online content itself is itself extracts or adaptations of such material.<p>What you note about physical places and being <i>committed to your mission</i> when that involves actually going to a place and being in the moment <i>is</i> valid, and is hard to replicate online --- networks obviate space, and records obviate time. But you can at least be cognizant of that and try to recreate the experience in some manner.<p>The <i>Getting Things Done</i> model of time-boxing, defining core tasks and objectives, and recognising that <i>you can only achieve a very finite set of tasks in a day</i> is useful.<p>There's feature of a newly-developed browser, Einkbro, that I've come to hugely appreciate: save page as ePub.<p>Unlike "print to PDF", Einkbro lets you save multiple articles as "chapters" of a single book. That book might be a day's reading, or a compilation on a specific topic. It recreates the affordances of <i>a bundled publication</i> as with a print newspaper or magazine, in which a single issue has a finite collection of articles which can be read, and when you're done, discarded. The Endless Scroll model of the Internet doesn't offer this. Compile your ePub, read it, highlight and/or take notes using your ePub reader. And when you're done, discard without regret.<p>(Other tools, such as Tree Style Tabs on Firefox, at least allow grouping of articles related to a given task.)<p>The first rule of research is to <i>define your goal or objective</i>:<p>- What are you trying to learn?<p>- What is a sufficient level of depth?<p>- What measures or indicators of quality are you looking for?<p>One of the tricks of literature research is to try to go directly to canonical sources. This is usually:<p>- Original books or articles on a topic.<p>- Textbooks or assigned reading from uni courses. Looking for course syllabi is tremendously useful.<p>- State of the art rarely proceeds especially rapidly, and even in the information technology / comp sci arena, landmark texts can be years or decades old. Present churn is most often self-promotion and product advertising rather than useful information. Another challenge is that many businesses treat support and documentation as monetisation flows (Red Hat, Oracle, ...), and choke off useful information. My take-away is to avoid their offerings entirely.<p>- Otherwise, official documentation is almost always preferable to alternatives, though there are exceptions with specific independent high-quality publishers. E.g., O'Reilly technical books of old (pre-2010), etc.<p>- Most technical knowledge tends to be far more modular and accumulative than is generally recognised. A 20- or even 40-year old Unix or C reference, plus a few specific updates on particular new developments, can remain surprisingly useful (Kernighan & Pike, <i>The UNIX Programming Environment</i>, Nemeth, Frisch). You <i>do</i> have to keep an eye out for dated information however. Even modern machine learning techniques largely date from the 1980s --- hardware made previous theory viable.<p>- If you <i>do</i> find yourself reading recent publications / web references, <i>look to see what sources they cite</i>. Those will most likely be the true canonical sources, most especially if several references point to the same sources.<p>- Beware <i>information</i> vs. <i>mythology</i>. Claims or recommendations made on the basis of tradition or present fads are <i>not</i> true information, and should be heavily discounted. This is a case in which more-recent (though not necessarily <i>current</i>) responses to canonical / original works may be of higher value.<p>References such as Mortimer Adler's <i>How to Read a Book</i> and several variants on how to read a scientific article can also be useful for both identifying useful sources and most effectively extracting useful information from them.