Reading in the computer age

Once upon a time, computers used to be used for computation. To some extent they still are. But for the millions of people with computers on their desks and in their bags and pockets, computers are more regularly used for communication. And no matter how much we might like video conferencing or VoIP, much of that communication is textual. People read and people write. A lot. I don’t have any definite statistics, but I think it’s fairly safe to say that the average person today writes and reads a lot more than they did ever before. While there has certainly been an increase in literacy levels over time due to a number of socio-economic reasons, technology has played no small part in the growth of human informational output (and input). Though there’s lots of ways to produce and distribute information, text is still the easiest and most popular and so it’s something I want to focus on.

World Literacy levels (from wikipedia)

World Literacy levels (from wikipedia)

With the rise of blogs and dead-easy online publishing tools, it’s become easier than ever for the common man to take part in an existing conversation or to start his own. Even more recently, microblogging tools like Twitter let you take textual snapshots of your life and start mini-conversations around them. Furthermore, social network sights let you create conversations around pretty much anything that you can find online (and a lot of what you find offline). FriendFeed and Facebook in particular do good jobs of encouraging conversations. While it’s become really easy to write and publish your thoughts (which I think is a very good thing), the first problem that creates is:

Who’s going to read all the stuff that’s out there?

Certainly not us. Let’s face it, there is far more text out there than any human could ever hope to read and comprehend in a reasonable amount of time. Being the tool-building primates that we are, the solution that we’ve started to come up with isn’t surprising: we have machines to do it for us. Search engines and the many recommendation tools are all essentially reading machines that go through all the text on the web and produce a very small subset that our limited human brains to handle. Sometimes the machines don’t quite do the job, so we need some human intervention a la Digg and Reddit.

But the question is still far from fully answered. Alex Payne of Twitter takes a long, hard look at the future of feed readers and comes to the conclusion that “feed readers as we’ve known them are dying” but neither algorithmic processing nor human-social-network filtering works well enough to be a viable replacement. I tend to agree and I don’t see a solution myself.

I prefer a judicious use of both computational and human filtering. I rely on Reddit and Hacker News to provide fresh, new information sources, and then use Google Reader or Diigo to collect, index and manage my readings. However, it’s not ideal and requires more manual work than I’m comfortable with. We will certainly need better semantic technology to make sense of the growing amount of data, thoughts and opinions out there. At the same time, it’s not just a question of finding and organizing information, but also of remembering it when we need to. This brings up Question 2.

How do I remember everything I’ve read?

I read a lot. I read books for school and pleasure. I read scholarly articles for research. I read blogs to get a continual stream of interesting ideas and information. I read articles that float to the top of social news sites. I read Twitter and Facebook to keep track of what my friends are doing (and sometimes to read things they’ve read and written). Even by a conservative estimate, I would say I read an average of 5000 words a day, probably more on some days. I’d be lying if I said I remembered even a fraction of this. I can remember perhaps the name of the article, maybe where I read it, a few keywords and phrases here and there. This isn’t just a case of bad memory, (though I’ll admit my memory’s not as good as it could be), part of it is a conscious decision not to hold in my mental RAM stuff I don’t need at the moment (or in the very near future).

It would be great if I could remember even detailed summaries of what I’d written, but in the absence of really good memory and recall, outsourcing my memory is the next best step. And it’s not just me. In the old days you’d have tons of bookmarks saved in your browser or maybe even save the HTML files to disk. I did a bit of both. Today I use social bookmarking in the form of Diigo and Zotero for managing journal articles PDFs. For any form of memory system, full search is a must. Loose organization is also essential. Dynamic, multiply redundant systems such as tags and keywords are superior to static filesystem-like organization. It would be nice if this tagging were smart and automatic, but even manual tagging is very useful. Highlighting and inserting notes is also helpful if used right and in proper amounts.

The main point of all these tools is to simply to make sure that you find what you’re looking for with minimal time and effort. In some ways, there’s a redistribution of effort to save-time from find-time in the form of tagging and the like. Some tools like DevonThink go a step further and automatically extract relationships between documents. Being able to highlight sections and add notes becomes useful if you’re reading articles as part of a research project. These tools aren’t perfect and certainly won’t do your thinking for you, but they are a step in the right direction.

Such tools only make sense if there’s really something you want to do with the information that you read. Thankfully there are a lot of things you could with it. You could (and probably do) share it with other people. You can also join the conversation surrounding any piece of text, either by commenting on it (if it supports comments) or posting the article to an aggregator like Reddit and commenting there. You could also use it as the base seed for your own independent piece. Personally I view all my reading as input that will ultimately get processed into some form of writing. In many cases it gets translated into blog posts, comments or email, but in many others its still waiting for the right time.

Writing in the 21st century deserves a blog post all its own. Since I’m already over a thousand words, my observations on writing are getting pushed to the next post. I want to explore the issues related to writing in a bit more depth because I feel they are a bit more nuanced (and as a writer myself, I have a serious stake in writing in the computer age). Stay tuned.

Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out /  Change )

Google photo

You are commenting using your Google account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s