Finding the Signal in the Noise: An Attention Gap Analysis

Jon Burke at Technology Review has written a short article that reminded me of something I've been meaning to write about for some time now:

"While most of these services claim they'll simplify your life by imposing some kind of order on your news consumption experience, we still wonder about their ultimate utility. It wasn't that long ago that simple RSS readers such as Bloglines, which allow you to subscribe to your favorite headline feeds, were hailed as the ideal method for organizing and consuming content. Call us old-fashioned, but we think RSS readers works pretty well."

I agree with John - I think RSS readers work pretty well too, however, they have their limitations: what they don't do well is to find the signal in all the noise - you need more than just a reader to do this even if all the feeds are defined by you.

I subscribe to people whose opinions and views I value, but when there is 'big news' (as per yesterday's Windows Live) or even 'little news', I want to do a number of things:

  1. know who is writing what about a topic
  2. navigate the discussions taking place (clustering / threading)
  3. find the smart thinking - valued by others
  4. define the topic I'm interested in
  5. filter by 'my A-list': by those whose opinion I value (that I'm subscribed to)

Not much to ask is it? 😉

We've still got a long way to go in the provision of software that achieves all of the above.  There is software and services that do parts of these well, but there is no 'holistic solution' (that I know of at least) that does all of the above...if there was I'd sign up (and if there is, let me know!)

I've created a quick table to illustrate the point. I've listed 4 different services that I think represent the current landscape and have compared their capabilities against the criteria defined above (#1-#5) - think of it as a rough gap analysis. I've also added a 'solution' column, suggesting what functionality/feature I think would be required to be added to fill the 'Attention' gap.


Finding the Signal in the Noise: An Attention Gap Analysis

Service Meets criteria #1-#5 Fails criteria: #1-5# Solution - filling the gap






Works well for 1#.

Specialises in # 2 and #3 as I could see who was writing about the topic that have been referred to by others (clustered as related threads)



Fails on #4 - It so happened that the topic I was interested in was being tracked - but this is not usually the case.  I can't define the topic I want to track, so doesn't do #4

Fails on #5 - somebody else is defining the 'A-list'.



Add a search function. Crawl all of the blogosphere (all of it, not just an 'A-list'). Problem here is that Memorandum requires human intervention as part of the editorial process, so can't scale up to meet this #4 requirement with current model.

If I could import my subscriptions (OPML file) into Memorandum I could view and filter against my OPML file (my 'A-list'), this could solve #5

FeedDemon (and most feedreaders)







Strictly speaking it did OK for #1

Works well for 1# and #4

Since all the content in the reader is defined by me, it meets #5.






Fails on #1 - (is intensely manual process: I need through each feed manually to see if anyone has written about a topic or article)

Fails #2

Fails #3, (unless I include the fact that the authors that I read are doing this for me, which I'm not)

#4 (unless I have pre-organised my feeds to match the topic - this is not practical in the context of following a particular news item/topic)

Add 'search my feeds' solves 1# and #4)

Add 'see clustered / threaded view' to solve #2.

Don't know how to solve #3 right now...







Does #1 as core functionality

Meets #3 partially using 'by authority'

Meets #4 requirement (has search, tags)


Fails on #2

Fails #3 as the authority algorithm tells us 'total' authority - doesn't take into account their authority on specific subject, or news item - very rigid and non-dynamic

Fails 4# as I can't define my 'A-list')


Add 'see clustered / threaded view' to solve #2.

Don't know how to solve #3 right now...

If I could import my subscriptions (OPML file) into Technorait so I could also view and filter against my OPML file (my 'A-list'), this could solve #4

 Windows Live


Meets #1 and #4 (can run search against my feeds)

Meets #5

Fails on #2 & #3



Add 'see  clustered / threaded view' to solve #2.

Don't know how to solve #3 right now...

I'm sure I've got stuff wrong / omissions / over complex solutions / additional criteria. Ideally I'd love to have this table up on a Wiki for others to add and edit. If you run a wiki that you think we could do this with, go for it and let me know - just let me have editing rights 🙂

As you can see I have a gap in my thinking right now about how #3 is solved in this context (other than the solution will be algorithmic)

Other thoughts are welcome, thanks.

Update: I've been corrected re: FeedDemon features (e.g. 'Watch' - works v.well) - so have updated table accordingly.


Tags: Attention, RSS, OPML

Comments (10)
  1. Nice post. Holistic solutions are hard, but I think if Memeoradum let users import an OPML file, that would be a major step forward. I think Ross Mayfield suggested this too. As far as I can see, Memeorandum must work on some kind of "list" in order to keep out splogs, but it could become broader by covering more topics.

    You’re right – Memeorandum lacks the depth you can get from your feedreader, and your feedreader lacks a threaded/clustered view to let you see the bigger picture. It can’t be long before someone ticks off all the points on your list, though.

  2. J Wynia says:

    What I’ve been doing to get past most of these limitations is by using an email account designated for this purpose only and using Thunderbird and Outlook’s views and saved searches as well as their just generally useful tools for sifting through the pile of information.

    My OPML file is parsed by a custom PHP script that grabs all of the new articles. It then grabs relevant information about each feed (from web service API’s, etc.) and each posting: keywords, relative popularity of the feed itself, any private tagging or trust rankings I add, etc. Those extra dimensions are added to each item as X-FeedKeywords, X-Technorati-Inbound, etc. The resulting feed item email is then deposited in the inbox.

    With the extra X-Headers, I can filter for all postings with keywords "microsoft" and "live", also tagged as "highly trusted" and posted within the last 24 hours. And, I can save it as a saved search in Thunderbird.

    I tried piles and piles of online aggregator tools and standalone tools, and was disappointed by them all. Ironically, this included the purpose-built features for RSS in Thunderbird.

    Almost without exception their only method of presenting me my feeds was in a straight tree structure with one folder per feed and highlighting unread.

    Overall, it meets all of the criteria you’re talking about.

    The calculated keywords and views/filters let you group by specific keywords or combinations of keywords. Email clients are built for threading, though my current system could be enhanced by looking for interlinking postings to change the "in response to" email headers that email clients use for threading. It allows completely arbitrary dimensions of meta data being attached to a given item, and mail clients like Thunderbird an handle filters on arbitrary headers.

    I set it up for myself because it leveraged the existing toolset. I have no doubt that eventually the RSS readers themselves will catch up, but the email clients are already there and I don’t want to wait. All that was necessary was to hook the RSS to IMAP email (for multi-pc synchro) and add the semantic analysis of each post as it comes through.

    I’ve been starting to use Outlook instead and am exploring the views it provides, but I haven’t been using Outlook2003 for long enough to quickly set it up to see if I’d like it. I do and now I’m figuring out how Outlook itself and basic scripting to enhance it can improve on the duct-tape setup I’ve built so far.

    I think the key for anyone looking to improve this space is to pretty much allow arbitrary views, sorting and grouping on a big pile instead of trying to store anything in neat folders from the beginning. Every single user is going to have slightly different ways they will want it organized. The tool that "wins" is the one that lets them do what they really want without a hassle. And, what they really want isn’t easily knowable in advance.

  3. MSDNArchive says:

    Thanks Peter…OPML import idea: glad we’re thinking along the same lines 🙂

    JWynia: Loved your comment…sounds like yuo’ve got it cracked…

    Can you do screencasts? It would be a great way of seeing what you mean – a couple of scenarios/ use cases would be fantastic.

    What do you think?


  4. Alex, I was able to follow the Microsoft Live chatter quite easily in FeedDemon by creating a "Watch" for these keywords/phrases:

    * live software

    * Windows Live

    * Office Live

    * Microsoft Live

    FeedDemon then watched every feed in my subscriptions and collected items containing any of these phrases.

    If you wanted a smaller result set, you could also create a watch which looked for both "Microsoft" and "Live" in titles only.

    And if you wanted to search outside of your subscriptions, you could add a "Keyword Search" which subscribed to a feed generated from a search engine such as Feedster or Google Blog Search.

    BTW, FeedDemon does have a "Search My Feeds" feature – look for the "Search" panel below your watches.

  5. MSDNArchive says:

    Thanks Nick, just played with it and works v.well! Will correct (that’s why I wanted this as a wiki – I knew I must have missed stuff).

    Funny, I’ve been using FeedDemon for ages and never actually tried the watch out. Classic.


  6. John Tropea says:

    In regards to point 2., I believe RSS Bandit and SharpReader do some sort of basic threading

    In regards to point 4. Feedster allows you to search within an OPML (generates a feed), but you can do this with Bloglines anyway…I suppose it’s good so others can do the same with your OPML.

    Your posts are on fire at the moment, look forward to reading your stuff!

  7. Technorati’s David Sifry has posted the latest of his regular updates providing a bunch of stats and…

  8. Wow, I missed this the first time around…

    Megite is going letting me do what I’ve been asking…

  9. 10 Random Thoughts.

    # 2 Repetition engines

    New media is rapidly becoming like old media


Comments are closed.

Skip to main content