Sometimes I feel like either the lone voice of reason in the wilderness or, alternatively, the one guy who just “doesn’t get it” with some of the more popular ‘net sites, but I have to say that the more I view Memeorandum, the more I dislike it.
It’s important to remind you that Larry Page’s big innovation in Web search was that pages that had more inbound links were the best, most authoritative matches for a given search, thus Google was born.
Memeorandum takes the logical next step, calculating the most “relevant” discussions in the Blogosphere by tracking a few thousand of the most popular weblogs in a given segment and analyzing what news articles or blog entries are receiving the most inbound links at any given time.
But the whole premise of the site is critically flawed.
The fundamental problem should be obvious: Google started us down this path with its turbocharged popularity contest where the page with the most votes wins, and now Memeorandum creator Gabe Rivera has added an amplifier to the echo-chamber of the blogosphere with his site, another popularity contest in the already skewed world of blogs and bloggers.
Memeorandum faces this problem at two levels: the decision of which weblogs to track (and yes, The Intuitive Life is on the list) and the core premise that the more bloggers who are pointing to an article, the more “relevant” or important it must be.
There just seems to be something wrong with this approach.
I keep thinking of the many, many times in life that the most popular, the most “relevant” (in this context, I believe they’re synonymous) isn’t the best. We can all name technological examples, but there are examples in every walk of life, from most popular restaurant (can you say “McDonalds”?) to most popular fashions (seen how pre-teen girls dress lately?), most popular politicians (’nuff said about that) to most popular books (versus those that are actually thoughtful and well written).
So would McDonald’s be the most “relevant” restaurant because more people “vote” for it (e.g., eat there) than a local bistro? I can’t imagine that anyone reading this would try to therefore suggest that McDonald’s is the best or most relevant restaurant in the world of gastronomy, even if it is the most popular?!
I will admit that trying to automatically identify the “best” or “most important” in a given segment is incredibly difficult, but an entire news analysis system based on popularity, on incoming links, isn’t a step in the right direction, it’s a step in the wrong direction: What I want to see are sites that help me identify those stories that I otherwise wouldn’t encounter through the regular media (both blogs and journalists).
But from the popular Dodge Interview with Gabe Rivera to the accolades from the likes of Pete Cashmore, Stowe Boyd, Om Malik and tireless promoter Robert Scoble, I’m unquestionably in the minority with my viewpoint here.
Is the underlying concept of Memeorandum — that the more inbound links a news article or blog entry garners, the more “relevant” it is to the community — a good idea or the latest incarnation of a popularity competition, with all the problems that suggests?
What’s your opinion?
Sometimes I like to cook my own meal using the fancy wok I got for Christmas and a made up recipe. Sometimes if I’m in a rush I’ll go to McDonald’s for a chicken sandwich. And sometimes I’ll go to a fancy restaurant with live music and a full wine cellar. I like having the choice.
While I understand your point, it’s important to remember the site is called MEME-orandum. A meme is “A unit of cultural information, such as a cultural practice or idea, that is transmitted verbally or by repeated action from one mind to another” (Answers.com). Thus, Memeorandum’s use of inbound links as a ranking tool is exactly right if its goal is to identify where the memes are — which, based on its title, is indeed what it’s supposed to do. For me, Memeorandum is just fine, but only when taken into context with all the other feeds I follow.
Using the concept of a “meme” doesn’t require it to be a popularity contest, though, Shel. For example, if you go back to Richard Dawkin’s discussion of memes, you’ll find that it’s the *transmission* of memes that’s key, not the popularity of a meme. How do you find the interesting and relevant in a sea of discussion?
Dennis, I agree with you about choice, but what sites do you use to find the interesting and relevant new ideas in the blogosphere, as opposed to simply the most referenced?
I think the algorithm, the principle is good. The results seem to vary with the sample, though.
I liked the original political sample, because it did expose me to new viewpoints, new ideas — events which were not covered in the commercial syndicators (AP, NYT etc) were viewable in the Memeorandum slice of politics. I could see which ideas one faction wanted to promote; which ideas another faction wanted to promote, and compare them.
The tech sample is still useful to me each day — it’s how I found your essay here — but the same algorithm doesn’t seem to uncover as much surprising news with this group. I suspect the social effects are different too — sometimes there seem to be calculated efforts to introduce a story to the front page. (Blogrolls helped networks establish search engine prominence a few years back, for instance.)
One thing I’ve wanted for a long time would be a clientside engine which does link-analysis upon my choice of categorized news sources, an agent which works in the background, and then which alerts me with a notification if a certain number of my trusted sources point to the same link, or if there are multiple links using certain hot body text (company or personal name, etc) — a configurable, personalized scanning of the news. But in the meantime, Memorandum does help me much each day, and I appreciate the work that Gabe has done for us all.
The 2008 US presidential election is coming up, and in spring 2006 candidates are already positioning themselves. The buzz for tech funding is another important driver. I think we’ll start seeing significant efforts to game these new “news” systems now. Rapid evolution ahead, now that the news aggregators have become the news….
Well all said and done I believe your concerns are well founded. For instance any political social group can dominate the headlines with a simple strategy of flooding links. The problem is that what is seen as most relevant for some isn’t always for the whole and of course there is the reverse of this.
What may be of importance to a particular tech segment may be of little importance to the other segments. What this particular methodology brings is highly biased relevancy. Maybe tweaking the algorithm would help by bringing other norms from outside the sphere of influence to better moderate and neutralize the bias.
In a business context this sucks because the most important issues aren’t the most popular issues. What is relevant to my field of Customer Service Development is of great importance to business success but isn’t necessarily high on the popularity list of most popular relevancy or maybe be over shadowed by a special interest sector that has a very twisted view of what is relevant in this field.
Most popular also may not be indicative of change but of maintaining the status quo such as is evident in the branding industry as the big firms and old school die hards resist the process of change to a customer based ideology.
I think you see where I’m going with this, which is pretty much in line with your view. Thanks for the piece.
I don’t like Memeorandum either, except as a big-picture overview of what’s popular. The problem with it in my view is that it doesn’t segment enough. Like you, I’m actually not that interested in “most popular” as a raw aggregate, but rather, what’s most popular in the segment of the population that I vibe with. I think that’s what a lot of big websites suffer from – trying to be all things to all people (same thing old-school marketing suffers from).
I’d like to see a memeorandum that tracked progressive business blogs, for example. Or a digg that was specific to self-employed business people.
I’ve spent much of today thinking about my posting and the thoughtful responses here, and I’m now trying to decide whether my dislike is muchly about the use of the word “relevant” in the description of Mememorandum, actually.
Popular just doesn’t equal relevant to me, and to say that it “uncovers the most relevant items” is to do a disservice both to the site itself and to the people who want to see what’s there. I’d still have concerns about popularity as a substitute for goodness, I’d definitely have less concern if Memeorandum described itself as “uncovering the most popular items” instead…
The sad fact is that so many people are really into popularity as their primary filter.
Sure, we can envision alternative approaches to filtering and prioritizing which adapt to each user’s profile, but there is probably only a very tiny audience of us who really do care about anything distinct from raw popularity.
Given a choice between truth and popularity, I suspect that most people will instantly (and instinctively) take popularity in a heartbeat.
So many people desperately want to be associated with what is “in” and “trendy”. They want what’s “hot”, not what’s not.
To your point, what is the meaning of the term “relevant”? Simple: In this particular context, something is relevant if it is relevant to *popularity*. See, relevance is actually a fairly neutral term, a kind of tool. You need to focus on the *object* of relevance, the “what”, and in the case of Google and its followers, the object of relevance *is* popularity, that which represents the direction that the herd is moving.
You (and I) and too few others would like a mechanism to *tell* these applications what our own preferences and priorities are, and then the applications will filter and order the results in a way that is *relevant* to *our* own personal interests. But even if all online applications did so, there is still the herd mentality that would want to “go with the crowd”, and go with a popularity vote of priorities.
Of course, you could also have a feature that allowed the user to select another user or group and request to see results filtered and ordered according to that other party’s interests.
The real problem here is that we need a lot more basic research into a lot of this stuff. People are putting these web sites up with far too little thought about the core issues. So many of them would barely even be acceptable as undergraduate term projects (or at least that would be the case if schools had reasonable standards of quality). Yes, give them an “A” for cuteness, but a “D-” (and “do over”) for failure to deliver meaningful, durable social value. And I’m being overly-generous with that “D-“. It proably should be an “F-” and the threat of expulsion.
Enough with all the web equivalents of pet rocks and hula hoops… show me some real value.
— Jack Krupansky
Back when the earth was still cooling and I was in graduate school studying information retrieval systems, we spent a lot of time discussing measures of “precision” versus “recall”. I don’t know if people still study these things since retrieval systems have changed so much, but I think the concepts are still useful.
“Precision” was a measure of the proportion of rerieved items that were relevant to your query. “Recall” was a measure of the proportion of items that were relevant to your query that were actually retrieved by the system.
Many papers were written on the mathematics behind these two concepts, but for me the relative value of the two measures always related back to the type of query you were asking the system to help you with. Were you, say, at the beginning of a search where you needed to cast a net as widely as possible? Or were you trying to locate a specific fact and needed to zero in on relevant information as specifically and as quickly as possible? Variables impacting your accomplishment of these goals included the completeness of the underlying database being searched and the sophistication of the search technology in converting your query into operation against the database.
Plus, what was the value of the time you could devote to the research? Could you afford to scan through a large number of retrieved but irrelevant items to locate the one or two items you actually needed? Or was it more important for the system to do the screening for you?
I think that measures of popularity have a time and place. I would never confuse “relevance” with “popularity” though since the terms are so relative. But I can conceive of many situations where knowing what a lot of people are talking about is an important and valuable thing to know.
(Dave, to answer your question — I use a variety of sources to find interesting and relevant ideas in the blogosphere. I have many stored searches run via RSS feeds for search terms I know that interst me — and I supplement this by coming to intersting sites — like yours — where I know I may run into something that interests me. )
There is one particularly good reason to find value in the service – if you are atop or in the middle of the echo chamber it creates, it acts as a good filter and link chaser. It is also a good way to join important conversations that people with wide distribution are having on topics that could use the input of others.
But memeorandum is really another view on a filtered feed reader. If you like the filter he applies, it is incredible. If you want what’s best instead of what’s popular, you need to create your own filter still…