mca blog

it's official! i've signed an agreement w/ O'Reilly Media to write a book focused on creating hypermedia clients. i've been putting this project off for a couple years and am glad i finally will be sitting down to cull through the collected material and work to put it into a (hopefully) useful presentation.

i've spent several weeks coding up lots of examples of hypermedia clients; experimenting with several existing media types and even some profiles. i've also been working through common patterns that make working w/ message-oriented services easier to handle for client apps. i have learned quite a bit over the last couple years of talking w/ people and assisting on various projects both large and small. and now it's time to see if i can find a coherent story in all that experience.

TBH, i'm not exactly sure where this one will lead, but i have a decent plan to follow before i start out toward the horizon. i won't belabor the details here but will pass along my general outline to give folks a feel for where i am trying to go...

Part 1 : Human-Driven Hypermedia Clients

i'll be keeping discussions of human-driven and (i'll say) machine-driven clients clearly separated in the book. not because there is always a clear line between then in real life (think parsers designed to prefetch content based on inline instructions, etc.) but because it makes the conversation a bit easier.

that said, the first part of the book will focus on the process of traveling from a client application that doesn't rely on any hypermedia, through several stages that end with a client that relies primarily on hypermedia information in responses. again, IRL there is a wide range of hypermedia in client apps for the Web (images anyone?) but this kind of approach will, i think, help bring some things into focus.

Part 2: Machine-Driven Hypermedia Clients

ahh, the fun stuff! the holy grail, the unicorn of song and story!

or not.

similar to part 1, i'll use the second half of the book to focus only on challenges related to creating hypermedia clients that don't rely directly on human intervention for each request. of course, humans will program them, but then -- at least for some level of interaction -- leave the machine to do their own thing. and that's where it gets fun.

i'll talk about how we can build a client that has the power to do basic interaction (independent request/response), how machines can be "taught" to safely navigate a selected part of the WWW environment, how they can recognize 'things' (affordances) along the way, and how to use those things as tools to manipulate the local environment. finally, i'll talk about how a client can be taught to recognize a 'goal' or 'end-state.'

in real life, humans do this kind of stuff every day. we navigate through traffic, recognize road signs/signals and are able to figure out when we reach our destination -- even if it is not the destination we had in mind at the start of our journey ("Sorry mate, the pub is closed. But the one up the road is still open."). we also go through these kind of scenarios in a more abstract or indirect way, such as when deciding we're done exercising, had enough to eat, or have collected enough trading cards to make a complete set.

it turns out, none of the cases i mentioned above require much "intelligence" or "reasoning." just the ability to pay attention to details and identify "done." and there are very many instances where this level of execution would be useful over the Web today. unfortunately, even these simple kinds of machine execution are not often supported when implementing services for the Web -- and that's a bummer.

but they could be. and that's cool.

On the Road Again

so, the next few months should be quite interesting. i always enjoy starting out on one of these adventures and this time is no different. i should also confess that i really enjoy getting to the end of a book project and putting the thing behind me ;) and i suspect that will be the true this time, too.

an added wrinkle is that i start traveling again in february and that usually takes a toll on deadlines. for that reason, i'll be drawing this project out a bit more. i hope to have the first draft completed by early summer and have it all wrapped up and out the door by fall 2015.

so, here i go, getting ready to navigate toward the horizon and hopefully recognizing when i get to "done."

should be exciting.

oops! road sign up ahead. gotta turn here. L8TR!