The 4 R’s: Search

Posted in technology, work by johnsnavely on December 7, 2008

Total Recall

So T delicioused me this article. It is thought provoking. And since delicious doesn’t allow for multiple comments or conversations, I’ll just have to discuss it here. This is going to continue an earlier post series.

We’ll try to concentrate searching and remembering, two tasks that we do all the time, offline and away from computers. But first I want to clarify a few things brought up in the article.

So here’s Philipp Keller’s primary polemic:

One common task while browsing the web is making sure you will be able to recall a valuable information you are just looking at. This article aims to prove that social bookmarking as in delicious, simpy, magnolia et al. is the wrong tool for that task.

I’m in “total” agreement. If you’re using delicious as your main tool for recall, you’re probably using the wrong tool. That said, is Keller seriously using delicious to remember stuff? He thinks delicious is the right tool for “Sharing Links” and Using “bookmarks to get things done”, but poor for remembering. Are those first two things really less important than recall?

Until I started using twitter, and the “share” option in my feed reader, my delicious feed was basically my micro-blog. I delicious stuff all the time with no clear intention of returning to it. Shit, I’ll delicious a link that I know someone else might be interested in just to have a conversation with them about it, even if I’ve never actually read anything on that link. (I may not have read Moby Dick, but I sure as hell delicioused it.)

Delicious is “social-bookmarking” or alternatively “url lifecasting” and as I’ve mentioned before it’s real power is in conversation and narrative. Delicious needs to buttress up these areas quickly or I’m going to export all my links to somewhere else and stream from there. Two features in particular drive me nutty:

1.) Why can’t I respond to someone’s description of a link? To have a conversation around a link that someone sent to me, I have to send it back with another “for:” attached.

2.) Why doesn’t a delicious post have a unique url? (This is such a pain in the butthole.) Then at least I could generate a twitter feed of all of my delicious posts or something like Feynman’s Turtles, delicious all the way down.

Granted, delicious also has poor tagging and search mechanisms. But even if these features somehow appeared in the next version of delicious, I wouldn’t think that it’s suddenly a memory aid. Why? Because I don’t want to “remember” by URL. Social bookmarking is great, but it’s not a perfect tool for “memory augmentation”. All it remembers are URL’s and tags; two fairly abstract methods of notation and organization. These only cover a small part of the things and methods by which I remember.

When you’re asked to search for the answer to a question, the question itself might fall into a range of categories. One extreme of that range is that the question is completely random, like trivia… like “who wrote ‘Fermata‘”, for example? Offline you’d probably go to a dictionary or encyclopedia. Online you’d probably go to your search engine of choice. This isn’t the same as remembering because you never knew the answer in the first place.

The other extreme is that you already know you’ve seen the answer, you just can’t remember where or how. The answer is buried somewhere in your collected detritus of bookmarks, files, emails, pdfs, or whatever. I would like to search my own stuff.

The category right in the middle is where I don’t know the answer, but I’m pretty sure my friends do, so I’d like to search through their stuff. Of course, these all exist on a range, so sometimes I might want to search a particular friend and their social network, etc. The closest thing I have to this is when feedly (which I like, but is still a seriously buggy work in progress) will show my feed reader matches for any google search that I do. I wish the delicious plugin did the same. But again, a search engine search doesn’t look at my “local” files, emails, or documents and when I need to remember something, it seems silly that I also have to remember the format of the content as well.

A tough recall scenario will go something like this. I need the answer that was a result of a conversation I had over the phone, that continued in twitter, that spanned a couple blog posts, that was mentioned obliquely in the description of a delicious url someone sent to me, that I annottated onto a pdf doc. These are remembered piecemeal, of course, which means I repeat a search for each bucket.

Keller thinks that one reason for the problem is because delicious tears links from context of the original page. This is 50% true. Contextual search/recall is what we need but the context isn’t just the page. The context is the series of thoughts and conversations that led me to the content, and also, where I think they fit in to what I already know. Why isn’t my all my data organized according to conversations or topics that I’m interested in? If it were, then I could grab things with even “less” context. For example, Internet Explorer webslices let me grab just a piece of a page. Sadly, MS tied these awkwardly to IE bookmarks. (I hope they get it right soon…those need to be feeds. Great idea, we hardly knew ye.) I would like to be able to grab slices of conversations, slices of videos, a piece of a song, or a section of a diagram. These snippets could be mashed-up into some other thoughts that might have little to do with their origin.

So what’s missing here? Well, I need a personal database or personal file system. (Live Mesh is an interesting start to the very basics of a personal file system. So far, I like it.) Then I can pipe all the pieces of my life stream (delicious, flickr, blogs, reader stats etc) as well as all of my emails and documents and conversations into a place where things I “know” are at hand and accessible. Then I can finally mine my own data. Which will allow for me to organize what I’ve seen but haven’t learned; an infinite stream of procrastination, ty CS18 & Professor Donald. Some of those promises I’ll make good on; others I never will. So having thousands of delicious links with no tags is fine. Not reading them is awesome. (Right now, I’ll usually only return to the most recent few hundred to find things anyway.) At the very least this personal data/file system lets me view my content in flows which match my life: a twitter comment sparked a delicious link sparked a blog post which was a conversation that I wrote a research paper about.

Anyway, there’s a lot more here to talk about. I can’t help but think of this recent obituary in the NY Times.


7 Responses

Subscribe to comments with RSS.

  1. son1 said, on December 8, 2008 at 4:13 pm

    I’m in “total” agreement. If you’re using delicious as your main tool for recall, you’re probably using the wrong tool. That said, is Keller seriously using delicious to remember stuff? He thinks delicious is the right tool for “Sharing Links” and Using “bookmarks to get things done”, but poor for remembering. Are those first two things really less important than recall?

    I had this idea last night, for a stupid little app that would take my delicious history and my browser history, subtract the second from the first, and then would (whenever I asked it to) randomly give me links to look at again. Maybe inversely-weighted by recency and the number of other delicious-users that have bookmarked it. Sort of a “reverse memory hole.”

    Actually, I’ve been thinking about doing this with my “starred” posts on Google Reader… whenever I need something to blog, I think I should just go to Google Reader, scroll all the way down to the bottom of the starred posts, and choose a link at random to write about.

    Anyway, and by the way, you should google the phrase “content-addressable memory,” and then let your mind wander (analogize) from there. There are some useful ideas in that phrase. (Some of the first machine learning applications of the things now called “graphical models” were meant for precisely this purpose. Some creative googling of the name Geoffrey Hinton would probably expand on that a bit…)

  2. son1 said, on December 8, 2008 at 4:17 pm

    Also, can I just say: Keller’s suggestion that Google search is much more effective than Delicious search rings exactly wrong for me. (Maybe he only tried to search on delicious before the upgrade? Searching before the upgrade was slow and severely broken…)

    The point is, I’ve been starting to consistently find that, when I’m searching for something about which I know absolutely nothing, Google does almost as good a job as Delicious.

    But when I’m trying to remember something that’s just on the tip of my tongue, something that I know I’ve seen before but I can’t remember what, Delicious is miles more effective than Big G.

    Although that might be a function of my own personal network. Maybe what Keller’s really saying is, “My network sucks, and I don’t even know it”?

  3. johnsnavely said, on December 8, 2008 at 5:17 pm

    Yeah, the comparison to Google Search threw me for a loop too.

    For “tip o’ the tongue” searches, I’ll use delicious/delicious network or search my feeds (which includes my delicious network as an rss feed) and in rare instances I’ll search my browse history, which in firefox isn’t all that bad.

    So that “stupid little app”… could you make that with pipes? Can’t you turn your browse history into an rss feed without much difficulty? (I’m just asking 🙂 ….)

    btw, thanks for the link. that’s another thing I wish I could do in delcious. give users mad propers for their link-fu.

  4. son1 said, on December 8, 2008 at 8:35 pm

    I mean, technically, since Pipes has that Web Service module now, anything could be solved with Pipes. I could solve 3SAT or the Traveling Salesman Problem “with Pipes,” if I wanted to go through the trouble of encoding the input as an RSS feed…

    But my Pipes-fu is rusty enough that I’m not going to try it anytime soon using only the Yahoo-given Operators…

    (We should page Bryan, and see if he wants to give it a shot.)

  5. johnsnavely said, on December 9, 2008 at 3:11 am

    well, the reason I want you to convert it to a feed is so I can listen in. 🙂

    delicious is getting crowded. I’d like an experience closer to when we’d work on separate projects, but in the same room. A small room, where there’s only a couple people in at a time, but we know in much greater detail what they’re doing at any point in time.

  6. jason said, on December 11, 2008 at 8:00 am

    Google Desktop Search does the search-my-local-files thing in a so-so manner… it has some hooks for a developer API also. Spotlight in OS X also tries, but it is tied to, which inherently limits it. It looks like a gdesktop plugin that also hits twitter/im/delicious/etc isn’t so hard, and it integrates with a search the same way you see with feedly (though honestly that 1.5 seconds feedly takes makes it unusable to me).

  7. johnsnavely said, on December 12, 2008 at 12:43 am

    well, let’s hack that shiz up.

    Then we can build whatever the real app is that comes after you got a somewhat unified search.

Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out /  Change )

Google+ photo

You are commenting using your Google+ account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )


Connecting to %s

%d bloggers like this: