This October I gave a presentation at Mashery's Business of APIs conference, speaking about how our API provides hyperlocal headlines on CNN.com as a case study for how APIs can power business relationships between tiny startups and multinational corporations.
The slides and (ack!)video from the conference are up, so you can check them out. In all honesty, I still haven't worked up the nerve to watch the video myself, so perhaps, my friend, you could watch it and leave me encouraging remarks in the comments? Please???
As a ruthless pragmatist, I'm always frustrated by the lack of practical take-aways from a conference, and the unconference is no exception. The strength of the unconference is that it accepts that you won't decide anything or make anything today and instead forces group contribution and constant socialization. Instead of listening to pre-appointed speakers, some people propose some topics, everyone shows up to the sessions that interest them, and you all have a nice dinner-table conversation about your topic for an hour.
This helps you get the two most important things you came to get: A) connections with industry peers and B) renewed energy about your industry. Note that A would be impossible for someone as pathologically shy as yours truly without the crucial forced socializaiton component. I've been to a regular tech conference before and I could barely bring myself to say hi to the fellow sitting next to me.
Some silliness I could've lived with out: writing "as a result of today…" on a post-it note and sharing it with the group (mine: "…I know how it feels to have a job where you talk to people all day"), a poem/rap/limerick-writing competition about what we got out of the summit, and RWW trying to sell us their $300 (not a typo) report on The Real-Time Web and Its Future for half-price. These felt a little like two sleep-away camp activities and a visit to the Scientology world headquarters, respectively. The bit where we got to give wine to people we thought did something good during the day was a nice touch and not overly sentimental.
Session 1: Truth-Detection on the Real-Time Web
I joined this session a little late. When I got there we were talking about the phenomenon of people thinking articles from The Onion are "real" news—partly because Baratunde Thurston was there. Shava Nerad pointed out that "The Onion and Jon Stewart aren't fake news" so much as a humorous commentary on what's going on in the world that "points to the real news" with the intention of interesting people in getting more information about what's going on in the world. I've been reading The Black Swan, so I'm not really sure what the "real news" is, but I agree that The Onion and The Daily Show are intended to be farcical commentary rather than misinformation. They throw a wrench in our problem because they sometimes unintentionally spread misinformation.
We discussed the problem of identifying bad information and tracking it back to the people who are spreading it. Any method of automating this would face the problem of distinguishing between those who are knowingly propagating the misinformation and those who are ignorantly repeating it.
At some point, we triedtostartrumorthatJustin Bieber got arrested. This was probably not a great topic because apparently this rumor had already been going on. A few people retweeted Baratunde's tweet, but I don't think anyone else in the room had followers who would retweet anything Justin Bieber-related. I know I don't.
We came back a few times to the idea of eBay-esque ratings systems for individuals' and organizations' reliability, but were perplexed by the challenges of people gaming the ratings for personal and political reasons. I asked: "Even if the system was working perfectly and my rating was a legitimate measure of how often I'd shared correct information in the past, how much confidence can that give you that I won't get bad information and innocently share it with you in the future?"
We decided that to get information out fast but maintain your integrity as an information provider, you have to be willing to correct yourself. We talked for awhile about the disproportionate sizing of articles and corrections in mainstream media. As Shava said, "Say you got it wrong louder [than your original bad info] and get appreciated for it."
Writing this up later, I wonder if even large retractions and corrections would be effective, because the original misinformation will probably have been reposted many times before the retraction is online. People who reposted the story may not check back with the original source hours or days or weeks later. Maybe there needs to be some inverse of a pingback system whereby the orignal source of a story can update repostings with breaking info.
At the end we talked about the importance of educating young people to think critically about the information they find and share online.
Session 2: Collaborative Knowledge
My friend and former outside.intern Cody Brown convened this session and kicked it off by mentioning a Wall Street Journal article (perhaps this one?) that described Wikipedia as a "crowdsourced" encyclopedia in such a way that Cody thought the term "crowdsourced" was pejorative. He also mentioned a blog post by Chris Dixon (definitely this one) that had posited that the most important startups in the past decade had been based on collective knowledge, citing the goog, the wikipedia, delicious, Yahoo! answers, and Yelp.
We discussed the advantage that aardvark and quora have over Yahoo! Answers of letting people know where their crowdsourced information is coming from. I somehow hadn't heard of quora but signed up immediately and am loving it. Whereas aardvark feels very invasive coming in through IM (which I hate with the fire of a thousand grandmothers) and never got my interests right enough to ask relevant questions, I have checked quora at least four times since I signed up on Friday and have found some extremely relevant questions that I really want to answer, such as: "How does outside.in get their traffic?" and "Why do some companies still force their employees to use IE6?".
We talked about what motivates people to contribute to collective knowledge and came up with two main buckets of motivation:
the super-user model, exemplified by (ma)gnlolia's "gardener" status, wherein people get privileges, influence, and recognition for contributing
the selfish motivation model, exemplified by bit.ly and delicious, wherein most users shorten links or save links for their own use, while unwittingly adding to a pool of knowledge about what URLs people are sharing and clicking on
Session 3: Semantic Analysis of Activity Stream Data
In this session we talked about the difficulties of doing semantic analysis on short status updates with a modicum of data to analyze and no standard taxonomy for presenting data. The only taxonomy that's been presented so far—hashtags—has been overrun with spam.
We didn't decide much in this session—the topic was a bit too specific and practical, and the number of attendees was a bit too small.
We discussed the tribulations of getting users to proactively add metadata to short status updates and the relatively small adoption rate of twitter location. A representative from TwitJobSearch mentioned that they crawl the links from Twitter profiles to get extra metadata about the tweets they analyze.
This session started out with four people in camera-less Section G (where, coincidentally, every session I participated in took place), but about 15 minutes in some folks from justin.tv came in and asked if we'd mind being moved to Section D, where sessions were being streamed live and, of course, recorded. The group quickly grew to six, then 10, then 15 people, with a few strays rotating in and out to see what all the streaming fuss was about, I suppose.
You can watch the video here if you want to see the whole thing, or check out my summary below the embed. I don't say too much—the other participants were pretty talkative—but if you're inclined to watch, there's a continuous shot of me alternating my best serious gaze between my co-participants and my computer whilst doing the following:
If you skipped out on the video, here are my notes from the session:
We started talking about foursquare and its privacy concern pretty quickly. Someone said that "foursquare is better at showing where you were than where you are," and we wondered if location becomes less important the less real-time it is. I pointed out (uh, rhetorically) that even if I had tweeted the latitude and longitude I had just shared privately with echo echo cofounder Nick Bicanic, I don't personally believe that my precise whereabouts at a single given moment make me particularly vulnerable. I didn't get to my rationale, but it's this: The cost of acting on real-time geographic information is extremely high. I don't think anyone wishes me ill that decisively.
We discussed the possibility of an "eBay for cabs" mobile app would allow you to share your location with cab drivers and find out how far away they are. Apparently such a one exists in San Francsico—it's called cabulous.
Bob Wyman—who had a lot to say on the subject—told us that his daughter carries an Android phone and uses Latitude to share her whereabouts with him so he doesn't necessarily have to call her and yell at her if she's stayed out too late. He also speculated that Abby Sunderland (the 16-year-old girl who went missing whilst sailing around the world alone) would've been a lot happier if she could've shared her precise location with people who were looking for her during her rescue mission. I wondered if she could've known that before her mast broke—making a solo trip around the world in a sailboat being of course one of the most brazenly independence-seeking things a 16-year-old girl might do—and congratulated Bob on having such a great relationship with his daughter that she surrenders her exact location to him at all times. I know my brother would've had part of no such thing as a teenager. Nic Luciano of GetGlue quipped "I'd be more likely to give a cab driver my lng and lat than my father." Ha!
Nick mentioned his feeling that the tendency to document our lives at every step—say, by checking in on foursquare as soon as we sit down at the table and tweeting a picture of our meal before we eat it—is a bit absurd in its interference with actually leading our lives. Bob countered by referencing a 1945 article from The Atlantic called "As We May Think" that suggests such documentation long before the age of "lifecasting" and "oversharing." I haven't read it yet, so don't spoil it for me in the comments, ya hear?
Personally, I found continuously tweeting pictures and observations from my trip to Ireland last year extremely helpful in reviewing and labeling with correct dates and locations the photos from my real camera after I got home. At the very least, we're making it easier to sort our photo albums and write our own histories by tracking our lives in real-time.
In Sum
I hope my notes help some people remember their sessions or participate vicariously. I'm looking forward to reading some other folks' writups. You can also check out the official ReadWriteWeb Photo Roundup from the event if you fancy.
If I was going to start a news business tomorrow, I would start a
news business designed to produce not one new bit of news, but instead
to aggregate news for individuals in ways that matter to them.
[I did tweet the slides in a somewhat timely manner, but that doesn't really count, now does it?]
Being pathologically shy, giving a talk to 60 people isn't exactly my idea of party time. A certain Kevin Prentiss scoured the internet for NYC-based product managers and conned me into agreeing to present. And I am ever-so-grateful to him for doing so—and for offering a brilliant last-minute reorganization of my presentation and some helpful tips for focusing my nervous energy. </3
Anyway. I think the presentation went pretty well considering that I was nearly hysterical with nerves. I even had a good time.
My slides are embedded below, and the notes on each slide, which I think you can only see if you actually go to slideshare, are packed with all the detail I went over IRL. Enjoy.
In other words, what's not to love? And I, for one, received a delightful Valentine's month surprise when I discovered that my broken heart/valid heart t-shirts are featured in not one, but two (!) Treasury Wests (Treasuries West?) on Etsy (both, alas, now expired):
Whee! Well, consider this your reminder to get a shirt (or, hell, a broken-hearted robot, it's pretty freakin' cute) for your favorite geek for Valentine's Day.
If you've been missing kenspeckly posts full of nerdy thoughts on the tubes, take a look-see at this post I just wrote for the outside.inblog about the ways Google Fast Flip misses the point in its attempt to innovate aggregation and UX of online news consumption.
The product claims to bring the experience of reading a magazine online, but the interface more closely resembles that of a microfiche machine (hat tip to Outiside.in Biz Dev VP Camilla Cho for the observation) and provides neither the physical immediacy of print nor standard web conventions to guide users through content.
We have this saying around the internetz, "links are the currency of the web." It's actually a bit of an understatement. Links transfer not only attention and authority, but also direction and meaning, helping us figure out where to go and what to expect when we get there.
Links don't just transfer meaning directly, when visitors read the context of a link before following it, but also indirectly, through the relevance algorithms of every major search engine, which almost universally include anchor text, even after Google's 2007 algorithm update to diffuse the most blatant link bombs. As Google's webmaster tools explains it:
[Anchor text is] how Googlebot sees your site. [...] This information provides good insight into how your site is seen by others.
This simultaneous transfer of authority and meaning is something I thought about a lot when writing a reeeeeaaalllly long paper on how search engines construct authority, and it comes up again and again professionally when the name of a project I'm working on doesn't correspond directly to search queries I want to optimize it for. Turns out Google rewards boring product names as much as boring headlines.
Anyway, enough with the 2-to-3-year-old search engine links. My point is that the skyrocketing adoption of Twitter and the increased use of URL shorteners it necessitates are altering the dynamics of power and meaning in the link economy. Anecdotally, I've stopped posting anything to my del.icio.us account with the intent of actively encouraging my friends to visit those links. I still save bookmarks in del.icio.us for refindability and tag clouds of my interests, but if I want people to look at something I'm thinking about or working on, Twitter is far and away my first choice. For a larger-scale example, consider that traffic to Fred Wilson's blog from Twitter has tripled in the past three months, and of course, the hockey-stick traffic chart of Twitter itself.
Embedded charts don't show up in most RSS readers. Go check out the original post.
It's worth noting that, according to bit.ly stats (which I obsess over), a reasonable portion of the traffic I send to a link via Twitter comes from it updating my Facebook status—and, indeed, half of that Fred Wilson post I pointed to earlier is about the increased traffic to his blog from Facebook. But Facebook's walled garden approach means that search engine spiders can't find the links we're sharing there at all, so it's pointless to spend time thinking about Facebook in this regard.
Links shared on Twitter, however, are almost always public to spiders as well as humans. Even though Twitter disappointingly barred Google juice from passing through its fingers by adding the nofollow attribute to all tweeted links back in September, most search engines do follow nofollow links—and do use the anchor text in nofollow links to determine the relevancy of a page for the keywords its linked from—they just don't use links with a nofollow attribute to calculate the ranking with which they weight the page's authority generally. But shortened URLs shared on Twitter can't transfer meaning to humans or to spiders because Twitter doesn't allow its users to create links with anchor text.
This can lead to dubious hilarity: URL shorteners make it way easier to rickroll—I mean, rickroll—your friends. The deeper meaning of rickrolling, such as it is, is that neither the clicking human nor the spidering robot knows what to expect on the other side of the link.
For humans who are not being rickrolled, the context of a Tweet usually tells them what to expect, but search engine bots are still at a loss for the meaning they usually retrieve from anchor links. This information is being collected by some URL shorteners—for instance, check out this bit.ly info page for a wikipedia article I recently tweeted—but it's not being passed on to search engines for use in determining relevancy.
This doesn't mean a doomsday for relevancy algorithms: Twitter may be starting to explode into the non-tech-elite world, but it's not outpacing anchor-link-friendly blogging yet.
If it does, I wonder how search engines will rework their relevancy algorithms to include "information on how [a] site is seen by others" in addition to the keywords its creators have planted on it.
Embedded charts don't show up in most RSS readers. Go check out the original post.
related: Check out this Hacker News thread on how URL shorteners are making money that was recently tweeted by my friend Aditya. Pretty interesting that there is no stunningly obvious business model for URL shorteners, despite their seeming positioning as brokers of the link economy. Those who are keeping thorough analytics on their short URLs (like bit.ly and cli.gs) are sitting on a gold mine of data, but it'll take serious work to monetize that information.
