‘Flow’ – day 7 – My Twitter thousands

Day 7 – The flow rises, but as it gets faster I just want more… I wonder what Scoble‘s flow is like…

Here are my previous flow entries so you’re up to speed:

Volcano Magma

I’m visiting my in-laws this weekend so haven’t been spending much time in the flow this weekend. However, even with short stints I’m finding a recurrent issue. Each day I think I’m going to hit a maximum number of people I can pay attention to. Each day I’m proven wrong. There’s an adaptation that takes place.

I’m following almost 3,400 and it’s working very well. I could imagine 5,000 being more than comfortable. Even on a standard IM client, the data flow is manageable. Most IM clients don’t smooth scroll, so it’s annoying to have each incoming tweet snap prior tweets upwards.

I’ve been thinking of the outline for a high-traffic Twitter client spec:

  • XMPP for tweet flow.
  • Web Services harnessed for contact management.
  • RSS/Atom integration for pulling articles from Twitterer.
  • Caching of existing Twitter contacts to embed information in to the XMPP traffic.
  • Search and real-time filtering.
  • Ability to only show tweets with links.
  • Additional filters based on: Age of Twitter of account, Location, number of tweets, ratio of following/friends, has non-default avatar, has non-default twitter design… with real-time color-coding of tweets.
  • Ability to favorite a tweet that came through XMPP.
  • Auto-pull of a Twitterer’s most recent blog entries (requires a scan for RSS feeds on the Twitterer’s home page, then pulling/parsing those items).
  • Auto-addition of Twitterer’s RSS in to Google Reader or other items.

With the above, one would have a complete Twitter news-room. One could immediately see what’s flowing and have access to a Twitterer’s additional information. This may be possible with a Flash or Java application, though I’d prefer a highly portable objective-C or C++ app. Maybe even ported to mobile clients (maybe, maybe).

This morning’s bugs with getting in to the ‘flow’ – starting day 5

Day 5 – 7:30AM EDT, 10 TPM (Tweets Per Minute)…

Links: Day 1Day 2Day 3Day 4http://twitter.com/solhttps://solyoung.com

Waterbug Faucet

This morning’s Twitter experience has found some bugs in the system. Last night I added a few hundred friends but didn’t go through my email for add announcements… Doing that now.

Most of the friends I added have added me back – I think making it clear I’m not a spammer and that I genuinely want to participate and learn from everyone in a flow helps here. There are some bugs with Twitter’s pages I’ve run in to.

As I go through my email I’m opening each person’s add announcement and visiting their Twitter page to confirm I’m following them. It would be great if the email described one’s own following status in relation to the new follower. As I visit a person’s page which I know I’m following, it shows the “Follow” button. Huh? When I click Follow, it immediately shows Updates as being on. My following count increases too.

Perhaps there is a difference between follow requests from a page and from the XMPP request? Maybe the AJAX request for following a person is getting bumped by the followup request to have notifications on?

In any case, I hope I’m not annoying people with multiple add emails this AM. If you’ve gotten more than one announcement from me, I’d be interested to hear about it.

As noted above, the flow is presently at 10 TPM. I expect the speed will increase by about 50% by 9:00AM. It’s pretty neat to see this kind of metric and have it readily available.

Twitter ‘Flow’ – Day 4 – Application Ideas and Metrics

Day 4 brings inspiration – this type of stream is like the Internet before Google…

A Flowing Pipe

It’s been 4 days with a flow approach to receiving data. I skipped adding more people today and focused on getting used to the incoming content. It’s become easy to follow along, so I’ll be adding again. Last night I experienced a reading nirvana while reading Robert Scoble and Shel Israel’s Naked Conversations (on the Amazon Kindle)… My reading was faster than ever. Unexpected and a real thrill.

The amount of data one can ingest seems like it could become a real measure of intelligence, like a hybrid or modified number of pages one can read per hour. Unlike pages, characters or kilobytes are easily measured and this type of ingestion stat could become interesting.

Today Twitter’s XMPP went offline for a couple hours. It was odd to not see movement out of the corner of my eye. Having the flow is no longer distracting (except when it’s not moving). I have it on the right side of my right hand 24″ monitor, and I scan it for links and more interesting items when I spend time on email (once an hour or so).

The metrics for the day with 2200 friends (averaged over a period of 10 minutes – after the jump):Continue Reading

‘Flow’ – day 3 – the volume is up

The flow is going and it’s time for plumbing improvements and deeper details on this process…

Really Big Pipe

Image courtesy of Komax Systems

Day 3

The question most people have been asking is, “What is the flow like?” Many have described this amount of flow as unmanageable and anti-social. Here’s what I’ve learned first-hand by Day 3…

After wrapping up yesterday’s post and promising to add 500+ friends per day, I destroyed my sleep cycle by obsessively discovering more than 1,000 new people. Since I’m a developer and VP of Engineering at iofy, I focused on developers and technology gurus. I’m also fond of the marketing and sales spaces as they relate to social networks, so spent some time beefing up that area of the flow too.

I do this by finding the most intelligent/witty/interesting people I can and spider through to their friends. Unlike a spammer, I only add a person if their tweets have been interesting and intelligent and I feel they’ll contribute to my education.

I woke this morning to a faster flow. At times today it closed in on my maximum reading speed, especially 9-5. With ~2,200 friends I’m now able to see instant changes in volume based on time of day, news, etc. Last night at 1:00am EDT, it was trickling. Before getting to the office it was still slow. Later, it drastically picked up. I’m getting metrics now and will share them tomorrow.

Our company president, @cart, supplied me with Steve Gillmor’s “Swarmtracking” this morning. Steve has a very similar approach but instead of using a Jabber client he uses the built in GMail web app and has search criteria. His article describes some good methods for tuning and searching, but the methods are distracting and require action (clicks). I also disagree with comparing this to a tracking system. One can use it that way, but it’s so much more powerful as a system for being fed valuable information.

What is the flow?

Reading and consuming the flow is like streaming a Google Search of the latest happenings that relate to you. Imagine a constant stream of somewhat relevant information. You scan as links and tidbits pass by. When something catches your eye, you click a link or respond with insight. Depending on one’s popularity, the flow splashes, much like a rock tossed in a river. One can see multiple splashes as multiple topics hit your flow at the same time.

The “Replies” page on Twitter.com works as an automatic net so I can listen to anyone speaking directly to me. It’s an automatic net and no further filtering is needed.

Unlike an RSS reader, this is real-time. My preference is to have an RSS reader open in 3/4 of my monitor and the flow open in the other 1/4. It’s immersion.

Additional thoughts and how-to (after the jump):

Continue Reading

‘Flow’ – day 2

It’s day two of discovering and opening up the flow… (not to be confused with ‘Flow Theory‘)

Flow - day 2

A couple days ago, after months of thinking about how to consume more information, I was inspired by Scoble’s post to switch off of a standard HTTP Twitter API polling application (Twitterific) and move to a Jabber based client (Adium). It has been an amazing discovery.

Initially, and until yesterday, I was using Twitterific to read posts from ~100 friends. I had SMS updates coming in to the iPhone for friends beyond the normal scope of Internet friendship (wife, co-workers, family). I bumped my friend count up to ~500 before my first flow entry last night

I picked these friends by viewing the friends of some of my other intelligent friends. If I found the last 20 posts from a friend in their list to be interesting and smart, I added that person as a friend of mine. If I found that friend to be exceptionally intelligent, I would review their friends and do the same process to find more. A tree diagram for contact spread would be very interesting!

500 friends created a slow flow in Adium (Jabber client). Today I followed the same process of friend finding and upped the count to 1,100. It seems intelligent people keep intelligent company (thank goodness!) and locating other intelligent Twitterers is not terribly difficult.

The flow speed at 1,100 is roughly 100 updates per 10 minutes (1 tweet per 6 seconds). Sometimes it gets much faster, but it’s easily manageable. With an approximate average of 100 characters per tweet and an average word length of 5 characters, this translates to 200 words per minute. At this point it is at a speed where one could read every post if they weren’t focused on other things, but more is tolerable.

I’m looking to have a flow that is well beyond fully readable. It’s supposed to be a river. I’m guessing this will be in the 5,000 to 10,000 friend range, but as I adapt it should grow. I’ll be growing my group of friends by at least 500 per day for the next X days to see how this works out.

I’m way beyond the point where I can pick out closely related friend’s tweets from the flow without software assistance. This also means it’s impossible to re-route the updates to a phone when away (my wife is happy about this ;). My solution has been to create a second account used only for following family and co-workers. My updates are still sent from the main sol account.

A latent side-effect of making all these new friends and finding all these smart people has been that they (likely, you) want to follow me too. Approximately 1/4 to 1/3 of the people I’ve followed return the favor and follow me. If you’re in to marketing don’t count on this lasting – I’m sure unscrupulous groups will use this against us and we’ll get a lot more careful in who we befriend.

For now, for those I’m connecting with, it’s a pleasure to meet you and thank you for making us all smarter.