Interview with Gabe Rivera, founder of Memeorandum

Tech Memeorandum is part of my daily reading list, actually I read it several times a day. There are several other "clipper" services out there that approach the problem from a slightly different angle. Blogs have exploded in growth and popularity over the last year. Finding good quality and popular content among the millions of blogs is the job of Memeorandum.

UPDATE: Since publishing this interview a new discussion on Meme Trackers has emerged. There are lots of start-ups out there trying to imitate or innovate on Memeorandum.

I had the chance to interview Gabe for this blog. Here is 10 Questions with Gabe Rivera, Founder of Memeorandum.

1. What were you doing before Memeorandum? Is this your first start-up?

Research on, and performance analysis of compiler optimizations, most

recently at Intel.  No web stuff.  First start-up.

2. When did you come up with the idea for Memeorandum? What inspired you? How long did it take you to build the first prototype, and ultimately bring it to market? How long have you been live?

Having watched the Jayson Blair debacle unfold on blogs, I became interested in the possibilities of blogging and wanted to try something new. Memeorandum 1.0 was kind of a "mainstream media meets political blogs" mashup deal, completed in spare time in about 6 weeks.  It went live in January 2004.  Example: http://www.memeorandum.com/04/04/01/  From this experience it occurred to me what I wanted in Memeorandum 2.0.  This took about a year to build, as it involved developing a very powerful article scraper, plus components for filtering and organizing posts and source discovery.  That went live September 2005.  Throughout, the prototype was the product, and vice versa.

3. How did you publicize Memeorandum and build up your user base?

I shared what I was working on with a number of bloggers before launch. Most were happy to write about it at launch.  Thanks Robert, Richard, etc! Since then, word of mouth/blog has continued to drive growth in readership.

4. How do you compare Memeorandum to Slashdot, Digg, Reddit, and others?

For readers of Digg (or Reddit, and to some extent, Slashdot), I'd say Memeorandum is:

- More focused (on either "Tech" or "Politics")

- More expert/authority-driven

- Better organized, visually

Of course for a certain type of reader, Digg's quirkiness, developer

orientation, and community are all pluses.  Many who aren't as interested in

these things prefer Memeorandum.  Many use both sites.  It's all good!

For the more ponderous set, I'd add that unlike all of those, Memeorandum

thrives on the web at large -- it doesn't keep its editors and content all

siloed and centralized.  Viva la edge, yadda, yadda.

5. How does Memeorandum decide what is important news? Some stories have no links, others have 10 or more links. How does content slide down the page over the course of a day?

Most-linked/most-discussed new stories are important.  Actually, all stories have been linked.  Items lack "Discussion" when people link without adding much commentary, since simple pointer posts usually aren't included on Memeorandum. Items fall down the page as they get older, or are displaced by bigger stories.

6. Reddit and Slashdot seem to use "expert/authority/reputation" systems to find and rank stories. Digg relies on its user community for "most linked/discussed" stories. These are two very different approaches. Memeorandum seems to use both approaches. I have seen stories on Memeorandum that are from expert sources with virtually no links or discussion, and other stories from lesser known bloggers that have several links. Does your algorithm measure both independently?

Yes.

(Editors Note: For competitive reasons Gabe's answer was short and sweet. The algorithms used are the secret sauce and highly confidential. Revealing more about them would allow competitors to copy them, and allow users to "game" them to optimize their posts for higher ranking.

7. When I worked at AltaVista we tried to measure how effective we were at producing quality relevant results by asking users to rank the results. We were never quite sure if they voted a result higher because it appeared on the first results page rather than the third page, or because they really thought it was the best result independent of position on the page. Does Memeorandum try to get around the "follow the herd" mentality? I guess this is another way of asking how do unknown blog writers get noticed on Memeorandum?

I think people are too quick to dismiss as a "herd" the many people who emerge to discuss a dominant topic.  Even when massive topic clusters form on Memeorandum, they often incorporate opposing or at least complementary viewpoints, with many of the more mundane posts reduced to a short "Discussion" links or omitted entirely. 

Though I like to see "unknown" writers featured on my site, I don't view it necessarily as a goal for Memeorandum to do so.  (Maybe "relative" or "former" unknowns is a better term.)  I want Memeorandum to interest its readers, and expose unknowns accordingly.  As a practical matter, Memeorandum can't make all or most of the unknowns happy.  It is, after all, a filter, and there are an awful lot of writers out there.

8. Are you surprised at the high quality of the content on blogs? How are blogs affecting traditional news sources?

Not surprised.  If I were, I wouldn't have bothered with Memeorandum!  I

think experts of various stripes are finding blogging rewarding, and I

expect this trend to accelerate.

I'd rather avoid wading too far into media punditry, but it's clear

traditional media ought to evolve in a number of ways to remain both

valuable to readers and profitable.  Among other things, I hope they go

deeper into blogging, making Memeorandum a better read.

9. What would you like Memeorandum to be in the next year or two? Something beyond Tech and Politics?

Yes, something beyond Tech and Politics.  In fact, I've already developed

demos in other topic areas, but am putting off launching until I'm happy

with their quality.

10. What is your business model going forward?

I think certain kinds of sponsorships will mesh well with Memeorandum.  More on this later this year.  There are some other services I'm considering, but it's too soon to blather on about them.

Gabe Rivera is a very busy guy. He puts an enormous amount of time, virtually every waking hour, into making Memeorandum the best possible service. My email exchanges with Gabe were typically at 3:00AM or later. He never sleeps...and neither does the news.

Ironically, this interview with Gabe Rivera may not show up on Tech Memeorandum. The algorithm for choosing stories is blind to author and subject. It looks for popularity (links out to popular stories) and authority (links in from authoritative sources) in ranking stories. Similar in some respects to search engine ranking algorithms. Like it or not, it is an excellent proxy for what the vast majority of users want to read. And that is what Memeorandum is all about...filtering through millions of blogs to find the nuggets users really want to read. Memeorandum has done a brilliant job!

Comments

[trickmcsneak]

Thanks for this interview Don. Somehow I haven't spent enough time at memorandum, but now I know a bit more about it - and how it works - I'll be there.

Thanks for the blog generally.

[Matthew]

Hi Don, you may want to look at Megite at http://www.megite.com too. Megite is the newspaper for anyone interested in what's happening right now by intelligently uncovering the most relevant items from thousands of news sites and weblogs.

[Ian Lamont]

Hello Don. Computerworld.com also offers a daily blog-monitoring service that combs IT-related blog posts for the best and most relevant entries. Unlike memeorandum, the posts are vetted by humans. It's called IT Blogwatch and is available at http://www.computerworld.com/blogs/?q=blogwatch

[Kevin Burton]

Well... it showed up just fine on TailRank :)

Kevin

[MonsterMan]

You might also want to check out http://www.newsbump.net . Nothing groundbreaking in terms of technology or innovation, just excellent content so far. Second only to memo methinks.

[dahowlett]

I'm not convinced with what Gabe says - whenever anyone asks the important question - "What determines selection/positioning on tech.me-me-orandum?' the answer is almost always evasive.

My sense in looking at the CODE is that Gabe selects what gets there and what doesn't. If I'm correct then this is a human algorithm - ie - Gabe selects what HE wants to see as the stories worthy of promotion.

Smart marketing but otherwise valueless to the wider community of people who look to blogs as a source of authority.

[Don Dodge]

Memeorandum must be evasive about how the algorithm works for two reasons; competition, and spammers.

If Gabe were to explain exactly how the ranking algorithm works his many competitors would copy it. There are already lots of competitiors in this space using various techniques.

Secondly, spammers, and even honest bloggers looking for higher recognition, would "optimize" their posts to "game" the Memeorandum algorithm.

Gabe does NOT do this personally or manually every 5 minutes. There is a very sophisticated algorithm that finds new content, aggregates similar content, then ages the content to slide down the page as it gets older and new stories overtake it.

I can tell you the algorithm takes into account the number of "links in" to your post, the number of "links out" from your post, and the "authority" or importance of both. Google's PageRank works in a similar way. In addition to the ranking algorithm there is another layer of algorithm that decides position on the front page based on "popularity" of the story and age of teh story.

If this were easy everyone would be doing it, including the big guys at Google, Yahoo, Microsoft, etc.

[Dennis Howlett]

Do they? Not convinced - as I've said elsewhere - I don't care HOW it works, all I want to know is the basis of selection. Content, topic or author - it really is that simple. The reason? Huge gobs of 'stuff' related to tech never get touched.