Prediction: As soon as Radio has tighter integration with Jabber and the benefits of IM (whether high level or low level like event notifications) become clear, IM will become a great thing again and only e-mail will remain as inferior. Until Radio does e-mail, of course, at which point e-mail will be good too. :-)

However, I think JR (wait! there are two of them! I mean John) is fumferring-about for understanding of the I/O medium. Context, fit-to-application, spirit and joy. Let's see how he does.

An ongoing question: How important is the number of writers?

If you recall IRC or public chat rooms, the doors are open to strangers, unlike with IM or private invite-only rooms. Controlling the doors, perhaps restricting poster population, improves relevance and quality.

When we talk of scale, lurkers are important to mailing lists and other many-many media. With topical changes and time, lurkers shift gears, becoming vocal, replenishing the stock of active posters. Counting lurkers separately is almost as important as counting posters.

I filter each mailing list into its own folder, and I set up rules to filter email from people I know into their own folders. The personal filters only apply to emails not already filtered by the higher-precedence mailing list filters.

The only email I see in my inbox is mail from people I rarely or never have received email from. And spam, but most email clients and servers have tools available to effectively filter most of that.

I use Nelson and Outlook to manage and navigate my mail.

I need smarter tools:

Detects clusters of similar ideas among different emails (e.g. here is a collection of mails from the KMnet, politech, and klogs mailing lists on a supreme court ruling).

Show me posters who think alike on about a given topic.

Distinguish personal correspondents vs. bulk mailers.

Learn to prioritize incoming mail by looking at my behavior, what I open, what I delete, what I read.

If this is so ineffective why not turn off all the lists and make everyone communicate via weblogs and instant outlines? Hey it's a bootstrap!

Reliable. Persistent. Pervasive. Democratic.

Don't walk away, build on it. Embrace it.

Look at the zaplet folks.

But you know what, email already does that, with the in-reply-to field, it's just that most email clients don't show your email in threaded form.

Side note: Conversant understands threads and shows conversations threaded on the web and through NNTP, even for communication that originates solely through email.

Side note 2: I think it's cool that Conversant sites/lists are called Conversations. They GET the importance of conversation in collaboration.

Like my email clients, it needs features that address the problems attendant with success. Volume of messages. Compound messages (like mailing list digests). Exploding number of contacts (see Ryze). Multiple languages. Threading, with splitting and merging threads. Categorization to help with search and retrieval.

Sounds like Google?

Among other things:

Please treat different objects differently:

• Time based events (meeting notices, television programming alerts, appointment confirmations, conference schedules) want to interop with my personal calendars (Outlook, Yahoo!, Palm, Nokia).
   
• Commerce transaction mail (buying, payment confirmations, banking statements, bills) wants to interop with my wallets (Quicken, ebanking, PayPal). 
   
• Career objects (CVs, resumes, job listings, interview appointments) want to blend in my jobspaces (job boards, candidate management systems, address books, CRM, resume subsites of personal web sites, bloggers for hire).

You get the idea. Mail is the medium; let's get to the messages.

Email is also an audit trail of sorts. If you sent the person the email, and it didn't bounce, they GOT the email (for all intents and purposes, anyway). That doesn't apply to pub-sub pull systems like email and weblogs.

From what I can see blogs and IO don't offer a particular advantage for scale.

The key is, are they appealing enough that authors and readers and conversationalists choose to use them? It has taken a lot of money, effort, and a decade to get half of all Americans to check email once a month. I'm not voting for one medium/technology or another, but the successor technologies must offer something so compelling that the average AOL Time Warner user changes behavior, adopts new habits, and demands their family, friends and collages join them.  

Weblogs have scaling issues in that the information you need in a one-off search is probably going to be surrounded by a lot of other information that is out of context and of no value to your immediate information need. Most weblogs also have inadequate localized search facilities. This can be remedied by having more advanced meta data on weblogs and weblog items, and having more weblogs for more discrete topics.

I didn't write this with the intention of mentioning Conversant a bunch of times, but I think it's worth mentioning that each "conversation" in Conversant can have as many weblogs (individual and/or group contributed) as desired. And conversations are thoroughly indexed and searchable.

Instant Outlines have huge scaling problems. Each time an outline is found to have been updated, the whole thing has to be downloaded again. If the updated information isn't at the top (or bottom in the case of chronological outlines) it's hard to tell what's changed since the last update. Downloading updates takes longer and longer as the outline gets longer (over time). The remedy is to trim the outline, requiring the author to archive their history. Weblogs already do this, they're inherently chronological and permalinked from the start. Instant Outlines don't have permalinks. Radio's outliner doesn't support the equivalent of HTML anchors, so you can only link to whole outlines, not particular outline nodes.

How does it rationalize? Which flow?

I'm guessing about rational, but I think John is referring to hierarchical rhetoric, the medium imposing a structure on thought. There has been excellent work on other forms: multimedia, hypertextual, narrative, recursive.

Scalability is a combination of technical and psychosocial factors.

Technical scalability, in the UserLand world, revolves around the limits of a server. Without caching, a low end manila server maxes out around 3000 active sites (ymmv). How do you become a Yahoo!, MSN, or AOLTW service for 4-5 orders of magnitude more users? UL chose to push cycles and bandwidth to the edge, to the users.

Scaling Search. Blogspace without robust search is mental onanism. As blogspace grows, search and retrieval becomes more expensive, challenging, and demanding.

Weblogs do lots of nice things for link-oriented search engines. Physically collecting data on a short list of topics (it's amazing how many sites have both cats and knowledge management); prolific cross-site linking and intra-site linking helps engines infer topic maps.

On the down side, weblog home pages are notorious for having too much on to many topics, making relevance much harder to determine, and changing focus faster than engines can believe.

IMHO, most of UserLand's dissing of Yahoo! and DMOZ stems from a combination of disdain/envy for the medium (after all, these are the web's most widely understood taxonomies) and the centralized labor-cost-per-post of a human edited directory. The only way to scale them is to add people faster than the web grows (fat chance).

Oh my! That applies to opml directories too!

Jim's notes on the advantages of blog over IO are real, but temporary. Implementation features vs. structural weaknesses. You can tell it is early days for OPML since we only have one vendor's edit/rend tool and the number of users can be counted on the fingers and toes of a millipede.

Psychosocial Scalability (my term) involves the limits of conversation and span of attention. The diminishing returns of adding one more person to a conversation. The limits to how much you can learn and retain in a given daily time budget.

Emergent properties of human behavior

The Instant Outline buddy window with it's bold lines for updated outlines is just as interrupt driven, and worse because you have no idea what has changed (no meta data from a subject line or similar) and no idea what's changed once you look unless it's plainly obvious.

John should also know from his Knowlege Management research that communication skills are learned and honed with experience, and sharing complex thoughts is challenging and time consuming no matter what medium is used.

If Instant Outlining was more like Paolo's Shared Outline system or the persistent threaded IM system I wrote over two years ago (in Radio), he wouldn't have these problems.

I use Mail.app on Mac OS X which indexes all my email and VERY quickly returns relevant messages to keywords in the Subject, Body, From and To fields of messages. Microsoft's Entourage e-mail client does the same and offers innovative "Mail Views" which are dynamic queries on its database of messages. I'd use it (I even paid for it) but it's database has a 2GB size limit which makes it unusable for me. I use Fire.app for IM which archives and indexes all my chat conversations. It also has a GUI for chronologically browsing and searching this information.

Conversant's threaded mailing lists / NNTP newsgroups / web discussion boards, fine grained weblogs, along with pervasive search technology solves nearly of the issues that John has brought up. Then again, so do best of breed e-mail clients and IM clients.

I believe greatly in the value of weblogs when they're done properly. Outlining is also something I believe in but it's application in collaboration requires some serious infrastructure which currently is not in place in Radio. Hopefully it will be some day. Until then, I see Instant Outlines as unorganized weblogs authored in an outliner.

Discuss...