Huffduffer

Editor's Notes

• #2: “Create a podcast of found sounds.”
• #3: I don’t mean “create audio files.”
• #4: A podcast is not an audio file.
• #5: Enclosures are a way of attaching multimedia files to RSS feeds (for download rather than embedding). A form of hypertext (hypermedia, really).
• #6: Jon Hicks created a podcast: an RSS file that uses enclosures to link to MP3 files on band sites.
• #7: Creating an RSS feed by hand is a pain. By creating a service to automate that process, I was scratching my own itch. If it’s useful to me, it might be useful to others too.
• #8: This is normally the point when a definition is quoted from Wikipedia. I’m going to quote Neal Stephenson instead.
• #9: huffduff.com was taken.
• #11: Handy tip for strict separation: develop as if for a multi-lingual site; it will stop you sneaking small “views” into your models or controllers.
• #13: Readable, guessable, hackable URLs are A Good Thing. Any page that has an RSS feed also has a JSON version (just swap /rss for /json). Also: xspf.
• #14: If you’re only going to get one thing right, get the typography right. Huffduffer conciously avoids rounded colours, gradients and so-called “friendly” typefaces. The look is sparse, slightly austere, even a little elitist. But that encourages thoughtful participation. Designed in the browser. Photoshop was used, but not very often.
• #15: Apart from the logotype, there is just one image on the site. This manicule is from the corner of Sackville Road and New Church Road in Hove.
• #16: An attempt to make the sign-up process suck less. This garnered a lot of attention. Dan Cederholm dubbed it the “mad libs” sign-up form. I had never heard of the game.
• #17: Not much JavaScript: just a bookmarklet to pop open a huffduffing window.
• #18: HTML5 just for the heck of it. The AUDIO element was dropped because of Safari’s aggressive pre-buffering.
• #19: XFN uses the rel attribute as a simple way of defining relationships e.g. friend, co-worker, sweetheart On the surface of it, the “me” value seems pointless. Actually, it’s enormously powerful, allowing you to associate multiple URLs with a single person.
• #20: Google’s Social Graph API makes the “elsewhere” list possible. hKit is used to parse hCards on other sites to extract the value of the “photo” property. In the case of Last.fm, it’s simpler to parse the HTML of a public profile rather than making an authenticated API call.
• #21: Coined by Edward Tufte: “data-intense, design-simple, word-sized graphics.” Made possible by the Google Chart API. Shows activity over time.
• #22: Also called triple tags. As seen on Flickr. Machine tagging *emerged* on Huffduffer. Xavier Roy (username: Jax) just started doing it one day. Machine tagging a file on Huffduffer is a way of initiating searches on other data providers.
• #23: The web doesn’t need another social network. The “for:” tag is taken straight from Delicious. iTunes + bookmarklet = a site you never have to visit. The ambitious plan: use the Social Graph API to drive recommendations based on relationships on other sites e.g. “You’re should listen to this audio file because it was huffduffed by this person that you know on Flickr, Twitter, etc.”
• #24: Huffduffer launched in October 2008. Becuase I’m a lone developer, a to-do list would probably be as good as using a bug-tracking service. One or two people asked for comments but nobody really uses them. Surprisingly, people aren’t clamouring for the ability to host files. The Twitter account is a bot. Either be a human or be a machine but don’t be both.
• #25: Building Huffduffer was a rewarding experience. Because I was scratching my own itch, even if no-one else used it, it’s still useful to me. As it turns out, there are now thousands of huffduffed files from over a thousand users. The next step is to use the network effects for recommendations and trends.