Problems I'm Interested In

I'm supposed to put this up on my personal site, but till I get Hyde set up and deployed I'm posting a list of problems I'm interested in (or thinking about) here.

The networked book. The recent industry-wide shift to ebooks is really no different from its physical analogue: you go to a store, you buy a book, you read the book. End of story. The networked book posits a radical idea: what happens if we could have a canonical representation of each book on the web?

Well, you can do cool things with that, of course. The obvious benefits are quick to present themselves: books in browsers means books are readable in the widest spectrum of devices possible (think: phones, OLPCs for kids in poor countries); you can aggregate reader-created comments and annotations into a central version of the book; you can link to specific paragraphs or sentences when referencing works on the web.

But the other benefits are not as clear - for instance, books can be updated if they're on the web. We can build versioning systems for books, and push versions down to ebooks as corrections are made available. Networked books can also be edited, and reading communities may (will?) spring up around them.

But also, the long-standing question: how do we get there?

Social software for small groups. The forum has proven — over the last couple of years — to be detrimental to the building and maintenance of healthy digital communities. Large social sites like Quora, Stack Overflow, Hacker News and Reddit have shown us that large online groups need tools to save themselves from themselves (and even smaller communities do better with self-enforcing controls built into the software). How would social software (e.g. a forum) would look like with these elements adapted for the small, digital community? The challenge is likely three-fold: a) how to promote good behaviour in online communities b) how to apply elements of gamification in the best possible way for maximum community engagement and c) the market problem - how to disrupt and replace existing bulletin board solutions.

Marginalia of the future. NISO (the National Information Standards Organization) has just started an effort to create a universal annotation format. There are multiple interesting challenges to this: i) how do you map annotations to specific paragraphs, given the fluidity of the text in most ebook readers ii) how would such a format be implemented, in order to be as cross-platform as is feasibly possible? (And what of PDFs? Scanned PDFs almost certainly cannot be annotated, as there's no text to latch on to.)

Note: there may be a good algorithmic approach to the first problem - e.g. using the Levenshtein distance to compare hashes of paragraphs, and this seems to be largely a reader implementation problem.

Open textbooks. The textbook industry is one of the worst industries to have blighted education for the past two decades (I'm not kidding about this, Feynmann has complained about it in his book Surely You're Joking, Mr Feynmann!, but the reality today is much worse).

I suspect that it's possible to disrupt this industry: provide free software for professors to write their textbooks in, on the agreement that so long as they keep a version of their textbook free and open (perhaps under the Creative Commons license), which we can then print and deploy to kids in poor nations.

Two quick things: 1) such initiatives already exist, such as in the state of California; but the software largely sucks; 2) Room for Reading seems like a good partner here, my friend Callie Miller is a big believer in what they're doing.

Can there be a better Wikipedia? Idle thinking; but I'd like to know if it's possible to build a better Wikipedia from both the software and the community perspective. The case for better software quickly presents itself: Mediawiki is rather obtuse to learn and to use (and is of course rather old). Also, are there better processes for topic review and deletion? I'm not so sure about that - the processes that exist today sprung up organically, and may very well be the best way to police Wikipedia.

A solution for the filter bubble problem. Eli Pariser has talked about this in his TED talk (see below); my question: is it possible to headfake a site — perhaps a news site, like Reddit — to solve the filter problem by purposefully throwing users up against people of different and opposing viewpoints? Albert Wenger of Union Square Ventures has suggested an opposite view reader, which sounds like an incredibly cool site to build.



Concurrent programming. I've not thought much about this problem, nor have I looked very deeply into it (will probably have to start reading papers, should I want to experiment in the field). But everyone in Computer Science knows that concurrent programming is the challenge in Programming Languages, at least for the next few decades. Liskov suggests that we need a new model of thinking about concurrent programming, the same way that Google's map-reduce approach turned out to be the right way to think about distributed computing. What might this mental model be?