6 search results for "tumblr search"

August 15th, 2008
April 8th, 2008

Track Tumblr, a Tumblr Search Engine

I created a little tool called Track Tumblr that let’s you track Tumblr posts for a particular search term.  If you’re familiar with the “track” feature in Twitter, this is similar to that, but for Tumblr. It presents search results in reverse chronological order that you can subscribe to as a feed and track mentions of the search term over time.  Say you worked for Starbucks and wanted to know what people on Tumblr are saying about your brand.  Enter “starbucks” in Track Tumblr, and you’ll see that Jakob Lodwick and other Tumblrs mentioned them recently in various Tumblr posts.  You can subscribe to the search results as a RSS feed or on iGoogle / My Yahoo!

Previously, the best way I knew to search across all Tumblr sites was to use Google and limit search results to the tumblr.com domain.  You do that by including “site:tumblr.com” in your search.  For example, if you wanted to find posts about Starbucks, you would enter “starbucks site:tumblr.com” into Google.  The problem with that is that it missed sites like mine that are hosted on a different domain name, joelaz.com in my case.  Track Tumblr attempts to address that shortcoming by leveraging FriendFeed search.  Many people like me have added their Tumblr feed to FriendFeed, which indexes the posts and offers a way to search just within the Tumblr service.  Track Tumblr combines Google Blog Search results with Tumblr search results from FriendFeed.  That’s why it was able to find Lodwick’s Starbucks post even though he’s not using a tumblr.com domain name.  The search results won’t cover ALL Tumblrs, but they will include any posts that are indexed by EITHER Google Blog Search or FriendFeed.  Hopefully, Tumblr with make this obsolete soon if & when they release their own search service.

Till then, I hope this is helpful.  Feedback is welcome.  

| |

January 30th, 2008

Tumblr API Feature Requests

I noticed a number of posts on my Tumblr Dashboard over the last few weeks regarding new features people would like added to the site.  The team working on Tumblr is faced with the challenge of addressing those requests while trying to maintain the simple, clean interface that drew many of us to Tumblr in the first place.  That’s not an easy balancing act.  David and Marco have done an excellent job managing the trade-offs so far.  For a future release, I think a good approach would be for Tumblr to expose some of the additional features through their API as tools for developers to customize their experience or to build services on top of the Tumblr platform.  For example, the latest release allows people with CSS / HTML experience to add a module to their page that automatically displays the people they follow on Tumblr.  Rather than impose that change on all users, Tumblr offered it as an option for customization.  That’s smart.  In just a few days, several developers created new themes that make use of the follower feature and made the code available to the rest of the community.  If you want to add a follower module to your page, you can.  If you think it’s clutter, leave your theme as is.  You choose the front-end experience you provide to your readers.  Everyone is happy.

I hope Tumblr uses a similar approach for some of the back-end data via their API.  If Tumblr continues to open up their data, external developers will meet the demand for new features without cluttering the Tumblr.com experience or tying up internal engineering resources.  Twitter, Flickr, and other sites have had great success with this approach.  Passionate developers build services that compliment those sites and address niche needs like a mobile interface, flash visualization tools, Facebook apps, and so on.  The current API is a great start, but it’s lacking some essential data needed from the developer community in my opinion.  

What would I like added to the API you ask?  Well, it turns out I made a little wish list:

  • Posts All: the posts “read” method should offer an option to query all Tumblr sites, not just one specific user.  User name / site name should become a filter option, but if you leave that field blank, the API would return a list of recent posts from all Tumblr users, similar to the public timeline on Twitter.
  • Post Search: keyword search API that returns posts matching a particular phrase for either one particular Tumblr user (site search) or across all Tumblr sites (Tumblr search).
  • Popularity Sort: the “read” method should offer options to rank results by some measure of popularity.  Popularity would be a function of the number of times a post was reblogged, the number of followers the author has, etc. 
  • Reblog Indicators: each item in the posts results should include reblog data similar to the info accessible through the theme tags (ie. reblogged from, originally posted by, and reblog count).
  • Following Method: a new API call that returns a list of all the people being followed by a particular user.  If people have privacy concerns, they should be able to opt-out of this feature and keep their follower list private.
  • User Method: a new API call that returns high-level profile data for a particular Tumblr user including their user name, site name, avatar, number of followers, number following, total number of posts, total times reblogged, etc.  
  • Groups Method: I’m not all that familiar with the groups feature in Tumblr, but I guess there should be an API call to capture that data.  

I’m confident that hundreds of interesting products and features would appear within a few weeks of exposing this level of information to the developer community.  I hope we see these API features added in an upcoming release.

If I missed any important API changes, reblog them or leave a comment.

Keep up the great work team Tumblr!

| |

January 14th, 2008

Discovery On The Web

Discovering personally relevant content online shouldn’t be so hard. 

I sometimes go to a bookstore with a particular novel in mind that I would like to purchase.  I’m focused on that task and my goal, in that situation, is usually to complete the transaction as quickly and efficiently as possible.  The online equivalent to this experience is a search on Google.  Search engines help me find things when I have a specific task or query in mind.  There is another side to uncovering content, though, that I believe is largely unsolved online today.  I’m working on building a company to assist in personalized content discovery for all sorts of content from across the web including articles, sites, videos, music, photos, and more.  

Most of the time I spend in bookstores isn’t focused on a particular task.  I’ll often walk into my neighborhood bookstore with no particular goal other than to discover something interesting.  I wander the aisles for a range of topics that interest me: new fiction, classics that I never got around to reading, books on architecture, beautiful coffee table books, or whatever else strikes my fancy.  Overtime, the store staff learns some of my interests and they occasionally make recommendations.  This is not a search experience.  It is not about finding a particular book.  Rather, it is about discovering things that I did not know to search for. 

Most of the sites I find myself using more and more lately address content discovery in some way.  I view my Tumblr Dashboard to discover interesting things from a limited group of like-minded strangers who I’ve added to my Tumblr follow list.  I visit Last.fm on a regular basis to view recommendations on new artists that match my particular tastes.  I subscribe to certain keywords from Delicious that are interesting to me.  I also use a number of vertical specific sites like Techmeme, which while not personalized, tend to focus on a particular topic like geek news.  

All of these sites are great, but the process for discovering content that appeals to me is painstakingly inefficient.  I shouldn’t have to read through hundreds of articles, sites, blogs, and emails to discover the one or two pieces of content that pique my interest.  I shouldn’t have to wait for a friend to send me that hilarious online video she stumbled across.  It takes me months, or in some cases years, to identify a handful of quality web sites on a particular subject.  It shouldn’t be like this.  I want a discovery engine.  I’m willing to divulge a lot of information about myself in exchange for personally relevant recommendations.  If a site like this existed, I would visit it everyday and I would be willing to pay for the service, either directly as a subscription fee or indirectly through some form of ads.  

I have a creative solution to this problem.  I’m creating a company focused on connecting people and their passions through personalized recommendations.  If you’re a talented engineer with interest in content discovery and recommendation technology, give me a holler.  

| |

December 29th, 2007

2007 ToKTA: Best Tumblr Meme

Aww. Rod Knowlton, Tumblr extraordinaire, awarded me one of his 2007 Tumblr awards:

FollowOnTumblr, from Joe Lazarus

With one simple run-on phrase, a list of tumblelogs he followed, and a polite request that everyone consider reblogging it, Joe made it as simple as a google search to find out what tumblelogs other tumblrs thought were worth following.

Considerably more work safe than the now non-existent Tumblr Radar ever was, and higher volume than the great Reblog Notes feature that was later released, FollowOnTumblr remains my favorite method for finding quality tumblelogs to check out.

(Image not visible in Dashboard)

Official 2007 Told or Known Tumblr Award Winner

toldorknown

I’m verklempt. Seriously, it’s an honor. Thanks man!

| |

November 27th, 2007

Tumblr Ping

Dear Tumblr, please fix your ping service so that my posts make it to Technorati, Google Blog Search, My Yahoo! and the rest of the blogosphere in a timely manner. 

XXOO,

Joe

| |