Scripting News

Another brick in the cloud

Thu, 06 Aug 2009 21:12:53 GMT

Got a couple of interesting calls this morning on the development of rssCloud. The idea is picking up steam. I'm liking it. For a lot of these things this is the second or third time I'm implementing them. That gives a certain confidence that you know how to do it, and know what the pitfalls are, and what challenges are coming down the pike.

One question comes up a lot -- how do you manage the global namespace in a loosely-coupled 140-character message network.

Twitter handles it simply, they have a server at twitter.com and when you give it a username it knows what data it applies to. It doesn't expose the internals. That's different from the Internet's domain name system, that turns a name like google.com, yahoo.com or us.gov into numbered addresses like 74.125.67.100, 209.131.36.159 and 209.251.180.18.

So in a loosely-coupled world how will this work?

1. We want to map names like jason, guy and carol to URLs like random.com/guy.xml etc, etc.

2. Probably the way it'll work is there will be a central server run by a foundation that does exactly this mapping and nothing more. It's an identity server. You sign up for a username and password and store one bit of data, the URL of your feed of 140-character messages.

3. Another possibility is to borrow the architecture of DNS, and make it a registrar problem. Sign up with Godaddy or Network Solutions or Gandi and create a domain. Now the challenge is to have that name point to a URL instead of a dotted ID. I'll leave that up to the DNS gurus to decide if it's possible or too egregious a hack.

4. Yet another possibility is to let twitter.com handle the bootstrap. Conveniently they have space for a URL in each user's profile. If you've got an aggregator and the user says they want to subscribe to someone named pearl, look on Twitter for such a user, get the URL from their profile, read it, and if it's an RSS file with a cloud element, you're home. If not, it's an error.

rssCloud news

Thu, 06 Aug 2009 13:42:48 GMT

The cloud keeps right on rollin along!

I love this piece by Rex Hammock on rebooting his Twitter follow list. It's great for a bunch of reasons. 1. Rex is a smart user. 2. He was inspired in part by Jay Rosen, who said in our July 27 podcast that he treats his subscription list as a resource for others who want to find people to follow in his field of expertise. 3. Rex relied on the rssCloud tool released earlier this week.

Developers -- it's our job is to listen to smart users like Jay and Rex, and then give them the tools to create the revolution. We are at best enablers. I do mean best. When we see ourselves as the show, we miss the point. The users are what's happening.

I started a mail list for developers working on rssCloud. Moderated at first to keep out the trolling and the spamming.

Now, that's not to say that developers can't be users too. The best ones are, and man that is powerful because there can be really good communication betw the user and the developer if they're in the same body. A good example -- Matt Mullenweg had the idea to import his Twitter subscription list into Google Reader and it worked. However, all the subscriptions were imported at the top level, meaning he had a clean-up to do.

So I added a feature to the app, if you add "&folder=1" to the end of the URL it creates an extra level in the OPML, designed for import into Google Reader (and probably other RSS aggregators as well). Example:

http://tw.opml.org/get?user=jkottke&folder=1

1. Save the list to your desktop.

2. In Google Reader, click on Settings (in the upper-right corner of the window), then Import/Export.

3. Click the Browse button and choose the file you saved above.

4. Click the Upload button.

5. When you return to the Google Reader main page, you'll see a new top-level section for your Twitter subscriptions.

Now you can follow Twitter folk in Google Reader. Heh.

Re Twitter's problems today. Centralized networks are especially vulnerable to DOS attacks. Loosely-coupled networks can do better. I wanted to post that to Twitter, but it's under attack. Not a joke, but something to continue to think about, planning for the future.

Tasted even better than it looks

Wed, 05 Aug 2009 23:13:35 GMT

Sampled at Sushi Ran in Sausalito.

Four stories

Wed, 05 Aug 2009 17:42:47 GMT

In 1997, the tech press knew three stories: 1. Apple is dead. 2. Microsoft is evil. 3. Java is the future.

In 2009, the stories are: 1. Heard a rumor. 2. Rewrite press release. 3. Steve Jobs. 4. Twitter has no way to make money.

So that's some progress. They added another story!

OPML for Twitter, Day 2

Wed, 05 Aug 2009 15:01:25 GMT

Great reception for yesterday's rollout of OPML for Twitter Subscription Lists.

Some thoughts from various people.

Kevin Tofel at jkOnTheRun asks if it makes sense to output not just the people you follow but the people who follow you. At first I was going to say no -- but then I remembered a very important thread that came up at the NYC meetup. What about when Ashton Kutcher wants to move his base from twitter.com to kutcher.com. He's going to need a way to export his follower list. So yes, it matters. Now I'm not sure I want to deal with an OPML file with over 3 million entries. So we need to think about this before writing the code.

Shea Bennett at Twittercism, a very smart guy when it comes to Twitter, has a twist on Kevin's idea. He says OPML can be used to move subscribers between Twitter accounts. Hadn't thought of it, but yes -- it certainly can.

Phil Torrone says MAKE is the only OPML magazine. He's into what I call reading lists. Same idea Jay Rosen came up with for his Twitter followers. I think this time around it's going to happen. I make a list of great people you should follow, and maintain it. When I add one to the list, you automatically follow. When I remove one, you automatically unfollow. My guess is the Twitter folks already have something here too.

As I said in NY, I'm wanting to build something alongside Twitter, not instead of Twitter. I don't believe in killing in tech, I believe in respectful and peaceful co-existence. It's what really happens. Not the drama the reporters want, but the continuity users want.

Anil's belly laugh

Tue, 04 Aug 2009 21:36:36 GMT

This should be fun.

An experiment. It's just a test. I put up a web service that returns a Twitter subscription list as OPML. You should be able to import it into your RSS aggregator or feed reader. But it's just a test. When the experiment is over the service will come down. It just builds on the Twitter API and took me a couple of evenings to put together. It's not Big Tech, it's just a little thing. But interesting? Perhaps.

For example, here's the subscription list for "cluelessnewbie," a fictious user who follows the 100 most-followed people on Twitter (that's what makes him so clueless).

Here's Jay Rosen's subscription list. Danny Sullivan. The Gillmor Gang. Gruber. Om. Loic. Doc Searls. Anil Dash.

When I sent a preview link to Anil, here's what he said: "This got one of those immediate belly laughs that only come from seeing something new and realizing exactly how disruptive it can be. :-)"

He's right -- it's part of Le Grand Bootstrap that's underway.

Who knows where it leads?

And that's why it's so much fun!

Here's the full writeup.

Chrome bookmark synch

Mon, 03 Aug 2009 23:58:53 GMT

I've been asked if Chrome's announced bookmark synch is enough to get me to switch from Firefox.

Short answer: No.

Why?

1. Right now all my machines run Firefox, and most of them are Macs. Chrome doesn't run on Macs. Yes I know they have a pre-alpha. Let me know when it's solid. And I might not do it even then.

2. I don't like it when a BigCo plays hardball with a little guy. I like XMarks. It works, and before Chrome competes with it, they should give them a chance to support their browser.

3. I like XMarks because a focused developer is going to care more about the service than a huge company that has lots of irons in lots of fires. They could easily forget about bookmark synch, if for example, a key engineer quits, or gets interested in something else. Bookmark synch is how XMarks' bread is buttered.

4. I'm scared of giving Google all my data. What if someday they decide I don't exist. What recourse would I have?

Rebooting the News #19

Mon, 03 Aug 2009 23:52:32 GMT

Show notes.

MP3.

Feed.

Followup on NYC rssCloud roadshow

Mon, 03 Aug 2009 17:38:12 GMT

Last Thursday we had a very successful meeting in NYC to discuss development of the 140-character loosely-coupled message network built on rssCloud.

The main immediate followup was to create an place for online discussion.

Let's use the comments on this post to begin.

My preference is to set up a YahooGroup, but if there's a consensus to use some other collaborative environment, I'll go with it.

Hey Mike, I told you so

Sat, 01 Aug 2009 18:10:18 GMT

In Sept 2008 Mike Arrington over on TechCrunch said that no matter how bizarre the setup over at the App Store, I would "keep developing for the iPhone," even though I had never developed for it. Obvious nonsense. Even if it was a wide-open platform I would have only considered investing in iPhone apps. I wouldn't have gotten the skill myself because it looked like a dead-end, and by Sept 2008 we knew it was headed toward the mess it has now become. You heard it here first.

The iPhone should have run the same software as the Macintosh. When I first heard about it, I misunderstood and thought I'd be able to write Frontier scripts that ran both on my desktop and the phone. I was a Blackberry user at the time, and I found the idea of a MacPhone truly inspiring. From there, it went downhill, and downhill and downhill. This platform was Apple's revenge on developers. Everything under their control. You couldn't even ship a product that Apple didn't approve of! Obviously that was going to be abused, and it has been, but finally it's become so ridiculous that it's obvious, even to Mike, that it can't work.

I've been through this loop many times, this is Mike's first. The only platform that really works is a platform with no platform vendor, and that's the Internet.

Steve Jobs is the anti-Internet. The Internet is utilitarian, it works, but it's ugly. Jobs's stuff is so beautiful that when taken to its logical conclusion, and he's almost there now, it's so dazzling, so beautiful that you fail to see that it is also useless.

There's still a ton of Woz in the Mac, I am typing this on a gorgeous unibody MacBook Pro, which is probably the most lovely computer I've ever used. The software I'm using has never been approved by Apple, and can be downloaded from the Internet. Next to it is an iPhone, which I use only as a phone, an IM device and a communicating camera. It sucks as a phone. The IM is okay and the camera is really nice. But as a platform it's a complete total disappointment.

The crazy ones

Sat, 01 Aug 2009 18:54:53 GMT

Bad Hair Day #6

Fri, 31 Jul 2009 21:30:48 GMT

This evening we recorded the sixth Bad Hair Day podcast. (not to be confused with actual BHDs which number in the tens of thousands)

You can find the show page on the BHD site.

The feed is handy for subscriptions.

And the MP3 if you're in a hurry and just want to listen.

From the school of hard knocks

Sat, 01 Aug 2009 03:44:32 GMT

Don't waste time on other people's qualities, intelligence, hypocrisy, honor. Distractions. What matters is what you do.

An hour with Jeff Jarvis

Wed, 29 Jul 2009 21:41:20 GMT

Just spent an hour at a Starbucks in NY with Jeff Jarvis of BuzzMachine. We talked about a lot of stuff, mostly about news and tech. Toward the end of the conversation he asked if I had any advice for his 17-year-old son Jake, who is already a successful software entrepreneur, having developed and sold a Facebook app for a fair amount of money, esp for such a young man.

Here's the advice I offered:

1. Be a user. Develop apps you yourself have a use for. If you don't have a feeling for what it's like to be a user, you'll never know how to evolve the products, and the stuff you learn in #2 will never make sense.

2. Listen to users. Learning how to code is straightforward, it takes time to perfect your skills, but it's relatively easy compared to the skill of listening. I recently suggested to a VC friend that we start a company whose sole differentiator is that it strives to perfect the art of listening to users. I feel that it would pretty much have the market of user-driven tech to itself. This doesn't mean that you get your drive from users, that has to come from your creativity, but it does mean you get your grounding from them. Without connecting with users, your products have no purpose.

Someday maybe I'll teach a course for software engineers on these two skills. It would be challenging, but out the other end would probably come a handful of really powerful entrepreneurial software people.

Did VoloMedia invent Podcasting?

Wed, 29 Jul 2009 16:41:02 GMT

They claim they did, in late 2003.

I'm certainly not a lawyer or an expert in patent law, but it seems the work Adam Curry and I did in creating the format and protocol for podcasting, in 2001, may have inspired their "invention." It certainly predates it.

1/11/01: Payloads for RSS.

Through out 2001 we did trials and experiments to learn how the protocol worked in practice. Radio UserLand, shipped in Jan 2002, was both a podcast distributor and a podcast client.

By July 2003, I had helped Chris Lydon boot up his series of podcast interviews with the new bloggers of the day.

All that happened before VoloMedia filed their patent application. Or so it seems.

ReadWriteWeb has a piece on the patent.

News from NY

Wed, 29 Jul 2009 00:40:22 GMT

I'm spending the week in NY with family.

Doing a rssCloud road show on Thurs and meeting with tech industry people on Thurs and Friday.

Yesterday Jay and I did Rebooting the News #18, show notes, MP3. Subscribe to the feed. Every week it's better than the last.

People keep asking for info on programming in the OPML Editor environment. I put together a list of resources. If you know of others please add a comment.

Went to the Mets game yesterday as the guest of Alan Levy of BlogTalkRadio, with my mom and Jesse Stay. I'd say it was the best Mets game I've ever been to. Fantastic come-from-behind victory where the deciding runs were scored with a pinch-hit grand slam. Alan really is up on the Mets so we understood all the strategy behind the Mets moves. Fantastic game. (Right up there with the exciting 16-run World Series blowout by the SF GIants of the Angels in 2002, which I went to with Scoble and Jake Savin.)

Saw this great sign on a walk today: Queens for Mike.

Two-way Search

Sun, 26 Jul 2009 18:41:38 GMT

When I started DaveNet in 1994 I had a bunch of ideas for products that I hoped one day to develop. But I had been waiting so long -- it was becoming apparent that I would never get to develop them. So I wrote emails about them, and sent them to all the people I knew from various industry events. What came back often were ideas that built on them. And eventually some of the products did get built. So the idea of dumping ideas publicly, ones that aren't doing you any good, is solid.

In that spirit, here's another -- I call it Two-way Search.

Here's the idea -- if the search engine knew a little about me, it could give more relevant answers. But it's too much trouble to enter demographic info, and I might not want to share that with the search engine company. But... There's a single piece of data that unlocks a vast trove of preference information -- the address of my weblog. From that it would be obvious that I live in the Bay Area and am involved in tech. So when I ask about New York style pizza, you might include places in Berkeley in the search results. When I search for a driver, I'm probably not looking for someone who drives a car. It goes on and on.

I call it two-way because like most things that show up on the Internet, at first search was a one-way thing. I ask questions, the search engine provides answers. By using information on my weblog to provide context, now data flows both ways.

BTW, it's conceivable that Google knows where my blog is, but I don't think they incorporate that knowledge in search results.

PS: Of course this is the solution to the Suggested Users List problem as well, if the new user has a blog. You don't need to know anything but the address of the blog to make intelligent non-random recommendations of people to follow. For one, you'd know what language the user speaks, so you wouldn't recommend 20 English-speaking celebs if they only know Portuguese. For example.

PPS: I thought the idea was so good I bought the domain.

Blogging at 35000 feet

Sun, 26 Jul 2009 22:44:25 GMT

Some day the novelty will wear off, but not yet.

Flying from SFO to JFK on AA 178. The picture was taken somewhere over Wyoming according to FlightTracker.

rssCloud design issue

Sat, 25 Jul 2009 18:19:52 GMT

Bryan Field-Elliot raises an interesting question in the comments on the rssCloud walkthrough.

"The automatic expiration after 25 hours seems rather arbitrary. If a cloud server is going to have a policy of expiring subscriptions after X hours, I suggest that the value X be published somewhere. Within the <cloud> element perhaps, or, as part of the return value from the pleaseNotify call. That way different implementations of cloud servers can vary this value as they feel appropriate, and different aggregators will have a means to know how often they need to resubscribe."

I responded as follows (and reposted here because I wanted to make sure it gets proper consideration):

It's an interesting question.

On the pro side -- it would add flexibility.

On the con side -- it would also add complexity. Another thing to test, another thing to break.

The subscriber should poll anyway, periodically. If it detects a change that it wasn't notified of, it can resub. No harm if already subbed. In my implementation of rssCloud that gives you another 25 hours.

I'm not sure I shouldn't give you another 25 hours anytime I detect that you're alive, for example, you respond to a notification.

The point is to give the cloud a rule for when it's okay to clean out garbage. It's really hard to imagine why someone would want a different value.

Loosely-coupled 140-char reading lists

Sat, 25 Jul 2009 07:00:00 GMT

I believe we will get beyond Twitter's very simplistic and limited Suggested User List, which I have written about so many times.

What sealed it for me was reading Glenn Greenwald's piece in Salon about Cheney's plan to use US troops inside the US.

Here was my thought process:

1. I'm going to pass this link on to my readers.

2. A very small number compared to the overall size of the Twitter base.

3. There are probably a few hundred thousand people who use Twitter who would want to know about this story, maybe 60K who would read it, as I did.

4. So how will this gap be filled? How will they find out about this story?

Well -- they won't.

But -- if there was a service they could subscribe to that alerted them to stories that would be of interest to them, based on their profile, a lot of people would give it a try.

In other words, I'm sure there's a place for editorial products delivered via the loosely-coupled 140-character network.

I tried this with NewsJunk, but it was either too early or there was something wrong with the way we did it.

It will happen.