1. 3
  1.  

  2. 2

    It’s called a ‘web browser.’

    1. 1

      Web browsers are awesome, but they’re designed for the exploration of content by individuals. The browser is an artifact of a world where humans are the primary direct consumers of content on the web. This is rapidly becoming untrue.

      A client able to communicate between any APIs on the internet can do very different things at a much larger scale. While I can use my browser to individually check online clothing retailers for cool jeans in my size, I’d much rather have my virtual agent do it for me, sending me a Facebook message when something new shows up. I’d definitely prefer if it could automatically incorporate new retailers without my intervention. We currently depend on third parties to do this sort of aggregation – Hipmunk, Google Shopping, etc – but this is basically an inefficiency.

      This is merely scratching the surface, too: imagine a client that could query for heuristics that indicate a global disease outbreak, alerts hospitals and automatically places pre-orders for antibiotics based on probable spread. Imagine if that client took 10 minutes for a non-developer health official to define. Try doing that with just a web browser.

      1. 3

        But he’s right. What you want is a hypermedia, so what you need is a medium (hypertext) and a protocol to transfer it (http) across a datagram network (ip+tcp) and therefore need a media representation and a client to represent it (html+browser). The combination is the world-wide-web.

        What you are critiquing is the notion that people don’t do it right. Which is true. It would be easier if people defined things a bit better. That’s just a question of evangelizing the now 50 year old idea that hypermedia is the goal, not the means.

        1. 2

          So is the browser in this model intended to provide intercommunication between unrelated web services without user input? The browsers available today don’t do anything like that.

          1. 2

            Yep. It is intended to do this. It reads html over http, which means it is the universal intercommunication thing you want. :)

            1. 1

              Since a modern web browser is about as close to this ideal form as a Cessna is to a spaceship, we probably have some work to do…

              1. 3

                I can take a web-browser and write some html and css and display it in some rectangular region of the screen. I write a script that uses the parser of webkit (a web browser, or more generally a hypermedia client, I guess I could use a javascript parser :P) to pull the html off of my twitter/rstat.us/github or whatever and produce new html (that I prefer) of my twitter timeline. I can inject avatars and links to github profiles using gravatar using a nifty little <img>, <object>, and <a> tags when I discover through a <link> tag on their rstat.us profile or something that they have an associated github profile. (which people don’t necessarily do, or do through something indirect like an xrd, but that’s kinda what you need to promote…) Hey, look at that, local hypermedia that involves linking to the global document view (world-wide-web) packaged in a way that eliminates user input and all you need is a universal client (web browser). :D

                Isn’t the past AWESOME? :P

                1. 2

                  Try doing all that as a non-developer. Job security is sweet, but I’d rather live in the future!

                  1. 3

                    I suppose IFTTTTTTTTx1000 is your thing then. You know what would make iftttttt actually good, and useful, and retains the crux of my argument? Fucking respect of hyperlinks. Like, it doesn’t look for rel=alternate for link tags to find out where things actually are, you still have to do the effort of finding the actual resource url for certain things, which I’ve complained at length about to their support team. That’s what /you/ want. You want people to use rels and links. And I agree wholeheartedly. And loudly.

                    1. 2

                      Yup. IFTTT shouldn’t need to exist :)