I had long expected the Yahoo Pipes end-of-life announcement, but it was still sad to finally read it. Wiring the Web, as Pipes and newer services like IFTTT and Zapier do, will continue to be a great way to get things done.
Last week, for example, I noticed traffic to a Web server where I'm running a soon-to-be-retired prototype of an Atom feed for our annotation service. The next day I found out it was a partner who's using Zapier to pull the data into a tracking system. Cool!
I've been poking around with IFTTT and Zapier too, using them to connect our Atom feeds to apps, like Slack, that in turn can send notifications to mobile devices. This is silly, simply a way to try out how it'll feel when we can do it more directly.
Is there an app for all the mobile devices, by the way, that cuts out the middleman and turns RSS or Atom feeds into notifications? If so I've yet to find it.
Some observations about the current state of the art in wiring the Web:
- RSS (and/or Atom) endpoints aren't as reliably available as I expected.
- The scope of the newer services narrows to pairwise connections among apps.
Neither constraint applied to Pipes back in the day, though, and it still never became an everyday tool for most people, nor do I expect that IFTTT or Zapier will. If you care about languages, tools, and protocols, you may think that people merely need a friendlier concoction of those ingredients than we've yet managed to brew -- which, of course, we do. But we may also need a different mental model for wiring the Web.
The Unix pipeline embodies the idea that apps can talk to each other, and you can make useful things happen by arranging conversations among them. When the apps are curl and tar, the outcome is downloaded and extracted data. But when the apps are Trello and Slack the outcome is messages transferred from one human context to another. Plugging one app into another to "configure an integration" doesn't seem to resonate.
Some of us are good at configuring integrations, and thanks to the new services more of us will be, but I don't expect that the Tinkertoy approach to wiring the Web will ever feel natural to most people. I'm starting to think that a better model might be a switchboard. It's still plugs and wires and sockets, but the connections are made between people not apps.
What would that look like? It's a bit hard to imagine, for me at least, because our connections to other people are mediated by apps and services that define communication channels and contexts. I don't make a single connection to Maryann or Jeremy, at best I can cobble together a set of connections via the Tinkertoy approach. To communicate I choose the tool first, then the recipient.
Another way, illustrated nicely by Windows Phone, is to pick a person first, then the tool. It makes perfect sense, and I thought I'd love that feature, but I hardly use it and I'm told that's typical.
Is the switchboard model wrong too? We can't really do the experiment because the services that I use to connect to Maryann and Jeremy regard them as products, not customers, and the clouds that deliver those services fragment their identities.
Will personal clouds that unify our identities, and coordinate our interactions with other people, make wiring the Web feel more natural? I imagine so but we won't know until, or unless, we have them.