lorennorman.com

Today, i’m going to the East Atlanta Beer Festival with my buddy Tom. I am not going to think about my computer, my company, blogging, Twitter, or video on the web for about 5 hours straight (1pm – 6pm) because I will be tasting some of 150 beers and having fun like the meat-puppets of the 20th century for once!

Here’s me talking about it:

And here’s what the action looked like last year:

Can’t wait!

TechCrunch just alerted me to this interview with the Twitter co-founders, Evan Williams (@ev) and Biz Stone (@biz), from this afternoon.  Here’s the video, and below are some quotes, notes, and my analysis:

A major strain on their system is power users. For example, people who are followed by tens of thousands of people, and avidly post (Robert Scoble.) The math doesn’t scale well at all, so when you overlap these “power-user” qualities, you clearly get an exponential explosion of activity. I can see how this is nasty to scale. Now imagine what’ll happen when the mainstream finally figures it out! (Hint: it’s happening. Have you seen @comcastcares?)

“Twitter could be down for months.” They’re discussing the fact that they have a plan for fixing their scaling issues, but that the plan won’t be in place tomorrow. They’re complaining that the users won’t accept “months” as an answer, but they really didn’t give any details. I’d really like to hear about their proposed architecture! Will they ::gasp:: shed Rails? Adopt the cloud?

“The ratio of downtime this week is better than last week.” Apparently Twitter has been undergoing immense growth. March was South by Southwest, which is where Twitter was born 2 years ago, meaning an enormous uptake of usage and users. In fact, according to Evan, users doubled in the March-April time frame. They made it through all of it without a crash, supposedly due to some recent optimizations. Of course, the growth has continued (i guarantee it is exponential and the rate is exponential) and further changes are required.

“‘Why don’t you drop Ruby already?’ That’s not exactly a solution…”

“We have a lot of money in the bank, money in the bank doesn’t make the servers run better.”

“We haven’t had serious talks about open-sourcing our software. We have a small group of smart people who haven’t solved this kind of problem before… We have some great advisers… Senior Engineers at Google… too many cooks in the kitchen [if we add more] at this point.”

“…we turn off the really expensive parts of the site so the site will stay up.”  Found that out the hard way recently, haven’t we?  I call them “brownouts”.

“API requests are 10 or 20 times as much traffic as web requests.”

Talking about a stream: “Lots of copies on cache, only one copy on disk.”  Sounds like a nightmare, but i’m sure he doesn’t mean “no backups” or anything so sinister.

“When we built Twitter, we didn’t know what it was.” This quote totally nails the power of Rails! I have to consider Twitter a Rails success.

Talking about what Twitter developers are capable of: “I can break this whole thing [Twitter], it’s fragile. It shouldn’t be, but it is.”  Way to be candid there, guys.

Why don’t you just stop taking new users? “That’s a possibility…”, but new users aren’t nearly as expensive as maturing users, basically, so it isn’t terribly effective.

“Twitter was a side project at Odeo.” 4 guys worked on it on the side “as a lark.” This is why they didn’t build a huge infrastructure up front. Frankly, it wouldn’t have gotten built at all if they had stopped to think about scalability.  (Rails definitely helped here.)

“Sometimes you have to build a prototype to get the point across.”  (Rails again.)

About spammers: “We look a lot at the ratios of users’ activity.” Apparently people are doing “traditional email spam” with Twitter by unfollowing and re-following to keep triggering the “You are being followed by SpammaJamma” emails, over and over again, i suppose in hopes you follow them back.

Any of my Tweetin’ audience have thoughts?

A couple of weeks ago, Adobe slipped this announcement past my radar screen. It seems we can download the Flash 10 Beta now (previously codenamed “Astro”), and it is chock full of new enhancements, most notably of which are access to Pixel Blender and 3D hardware acceleration.

Pixel Blender is a technology Adobe rolled up for After Effects CS3, originally. It basically gives the developer a lot of easy access to image (bitmap or vector) and video manipulation functions. If you are a hardcore type and you download Flash 10 Beta, then you should check out the Pixel Blender Exchange for a lot of demos showing you what things are possible with it. Here’s a video to show some of it off, as well:

Now let’s talk about the hardware acceleration. This should be a great addition that makes everything just gorgeous, right? Actually, no. It seems that it actually doesn’t speed up your traditionally optimized Flash apps, in fact it slows them down.

Does that seem counter-intuitive? Maybe it is, but it’s no real surprise to most game developers. 3D graphics in general is hard stuff, and as a community we’ve had to invest major amounts of time and money iterating our skillsets, toolsets, platforms, and teams to get to the point where we are today. This will also be the case with Flash 10, though the timeframe will be massively compressed (think 1 year instead of 10.)

There are already some amazing 3D libraries for Flash that actually work really well for us hardcore developers. But Adobe’s press release says that those options are “reserved for expert users” and that Flash 10 will make the functionality “available to everyone”. So does that mean that these frameworks will be able to actually just bootstrap the new hardware acceleration and get better seamlessly? Or is Adobe posturing this technology to compete and cause dreaded fragmentation, here? Eventually i think the point will still be moot, though (in a year the tools will be settled, regardless.)

As the confused juggernaut of Microsoft approaches this space with its usual lumbering pace and lack of direction or solidarity with Silverlight, it looks like Adobe is taking the appropriate steps to keep their 98% slice of Flash-enabled web browsers fully geared up, and its developers fully tooled up. Expect these improvements to trickle down to Adobe AIR rather quickly, as well. And if Google’s announcement yesterday is any indicator, i’m sure they’ll be keen to get some 3D acceleration in the browser as well…

Special thanks to Nick for turning me onto this bit of news.

Joel Spolsky (of Joel on Software) and Jeff Atwood (of Coding Horror) have created an excellent podcast for programmers called StackOverflow.  Listeners can submit questions by emailing MP3s, so i promptly sat down and recorded one. But i think it’s silly to just send it into the ether and wait for someone else’s users to one day get a hold of it and begin discussing it, when i’ve got a perfectly good community right here! (all 3 of you…)

So give a listen and say something productive, you heathens:

Here’s the transcript for lazy listeners, Google spiders, and the Cubicly Challenged:

====================================

Hey Jeff and Joel, great podcast.

My name is Loren Norman, and i’m a web entrepreneur and Ruby on Rails programmer in Atlanta, GA.

Joel, for your RailsConf talk, i think you should address the subject of Ruby in the enterprise.  There’s a bit of a holy war going on for these fundamental Rubyists who shout “Ruby is so great, everyone should use it.”  Of course, no language or technology can inherently belong, or not belong, in the enterprise, but the fact is it’s just not instantly practical for a corporation with millions of dollars of infrastructure in Java or .NET to suddenly inject Ruby into the mess.

So i think the real question is more along the lines of “How should a new language or technology go about entering the enterprise?”, is there a responsible path? perhaps it’s a task-by-task judgment?  or maybe there ARE there some things that should inherently be true of a language before it should be considered?

So what are your thoughts, guys?

====================================

And that question goes to my readers as well.  To the comments!

(Note: Marty’s coverage of this podcast here.)

Another podcast! Alas, the program remains nameless, and there has been no meaningful feedback. We forged ahead nonetheless, and now we’ve created even more discussion around cloud computing and live video broadcast to the web. Give a listen:

Topics:

• Loren is considering alternatives to Justin.tv for his live video needs. Does anyone know anything about Stickam or Ustream?
• Google FriendConnect: Webmaster’s Godsend, or Evil Empire Play? Facebook certainly doesn’t like it…
• Loren broadcast his entire day yesterday, including home-brewing some beer, playing some old-school video games, a health round of drinking games, and then some chick shenanigans
• Loren also likes to embed a Backnoise channel under his live videos for easy chat support/interaction
• Marty spent a long time trying to get Canadian sports with web hackery, and then discovered via a local blog, The Burgh Report, that Justin.tv is full of people (illegally) rebroadcasting pretty much all sports
• Marty discusses at length his experiences with cloud technologies, particularly Google AppEngine and how badly they have hobbled the excellent Django web development framework for Python
• Lastly, the guys discussed the new GPS, connected, crowd-sourced mapping device, the Dash, and how cool it is that it was built on the open cell network framework, OpenMoko

Now gimme some comments!

Today I am attending Startup Riot, an event for startups who want to pitch themselves to venture capital and angel investor types. There’s something like 50 startups presenting today, so they each get 3 minutes and 4 slides. It is going to be brutal. Chad and I are here just to watch the bloodbath and do some networking.

UPDATE ABOUT THE VIDEO STREAM: Well, I was broadcasting this event live before lunch, but I was asked to stop by the organizer. (Something about some people still have day jobs, potential SEC trouble, etc. I’m sure i wouldn’t understand…)

UPDATE 2:  Here’s the videos i took until my battery died, neatly archived by Justin.tv.  I’m linking them here for posterity, but they really aren’t that interesting when they’re not live…

Update: Marty has posted his coverage of this podcast here.

Saturday afternoon, Marty and I ground out our first podcast. It is literally me and him chatting over Skype for 45 minutes. Here’s the podcast, the summary is below:

Topics we cover:

• Podcasting software: why are there so many pay walls around letting me record my own audio? Currently we are using Pamela, a Skype plugin that outputs mp3s
• What should our podcast be called? What should it be about? Currently we’re thinking development talk, social web talk, and entrepreneurial talk.
• Should Marty write a podcasting hub on Google AppEngine + Amazon S3?
• “Flawed is the New Professional”
• Loren discusses his experience as an “expert” on web video at the Entrepreneurs Meetup
• CamTwist is amazing, free video broadcast software for the Mac
• Marty and Amy’s Credible Seminar Series

Why the hell did we do this?

Basically, Marty and I talk about a lot of topics via instant message on a daily basis, and it was time to start eating our own dogfood. We figured, if we’re going to have the discussion anyway, why not do it in the easiest way possible (talking instead of typing), and why not record it and give it away (podcasting to the rescue!) Of course, we learned a lot along the way just getting this first one completed.

Primarily, we’re unhappy with the fact that MOST people can’t just decide to start a podcast trivially, the way we did. And, of course, even we had plenty of bumps. A lot of this discussion is about these difficulties, the type of podcasting system we’d like to see, and how it can be really meaningful, socially.

Marty has actually begun work with Google’s AppEngine and Amazon S3, with hopes of getting some kind of podcasting hosting system created that we can use. If that happens, we’ll likely begin to host this podcast on its own domain, and set it up for subscription via your favorite software (iTunes or whatever.) For now, just look for future posts on our respective blogs.

Speaking of future podcasts, we need feedback and support! If you are interested in more podcasts like this, please leave me some comments here. I want to know what you liked and what you didn’t. I want to know what you would name this podcast. I want to know if you want to be a guest host with us in the future.

Thanks for listening!