Dec. 13, 2006, SAN FRANCISCO, CA — An unexpected outage at social news site Reddit today is having unintended consequences. At tech companies around the world, morning productivity has surged as geeks habituated to a morning of reddit-inspired web browsing find themselves utterly undistracted, with nothing to do but their jobs.

At Renesys, a company that delivers Internet Intelligence data for competitive intelligence, a hush has fallen over the virtual office. Normally, an IRC (chat) session used by employees echoes reddit, as individuals arrive for work and– choosing the instant reward of procrastination over advancing their personal and corporate success– call out links to wacky videos, stories of technical feats, and pictures of a VW Beetle with a jet engine. Without the fodder of reddit, however, this gang of introverts is compelled to speak mostly of their work.

And how little there is to say. “There’s a big list of links on the bug tracker front page,” said programmer Don Weiden. “I guess I’ll middle-click all the interesting ones so they open in tabs… heh, here’s a good one: clicking ‘Cancel’ causes a core dump. Oh shit, I programmed that. Excuse me for an hour.”

Asked if a competitor news site such as Slashdot or Digg could supplant reddit today, reactions were mixed. “I can’t wait 10 to 14 seconds for a page to load,” said Renesys COO Todd Underwood about Digg, which has a reputation of suffering poor performance due to its popularity. “I’d rather spend 90 seconds bitching about it in IRC.” For software developer BJ Premore, however, the downtime is scarcely an issue. “In all honestly, I check reddit every day or two, and hadn’t noticed the outage yet,” he wrote. BJ is regarded as one of Renesys’ most efficient workers.

EMPLOYEES SHARE THEIR LOSS

United by the disappointment of repeated, unsuccessful reload attempts, workers attempted to cope with their loss by analyzing it. “Look at the dns right now,” typed Underwood. “There are not even any NS records for the reddit.com zone on the servers pointed to by the authoritative .com servers. It’s going to be a long day.”

“I’m shaking a little,” added Premore.

Though reddit continues to be inaccessible as noon approaches, workers are
confident a solution will be found. “In the worst case, it wouldn’t take long to program a brand new reddit,” said Weiden. “I mean, I’ve been dying for an excuse to learn Erlang.”

Renesys is a real company where this blog’s author works. Most– but not all– of the employees quoted here are real, too.

Maybe this would be a good project for DevHouse Boston.

You have a cell phone. But you’re visiting family and don’t have an alarm clock with you. Why not schedule your wake-up calls via the phone?

Or suppose you remember you’ve got to start preparing lunch at 12:45. Why not have an automated reminder service call you?

This is just about the simplest possible VoiceXML service, but it’s still not done well on the web. The closest thing I’ve found is ifbyphone.com, which has basically the right services and business model but an interface that’s about 50% more cumbersome than it needs to be. They’ve taken the approach of offering as many options as possible, and thus it takes a while to find the free 10-day trial. When you call their number to schedule a reminder you have to wait through the greeting, say “wake up calls” (probably more than once), and then choose the day (say “Tuesday” or dial 1, say “Wednesday” or dial 2, etc… you can’t just say “today” or “tomorrow”) and enter or say the time. Then it asks you for your time zone, even though you entered a time zone during website registration. And finally it requires you to choose the phone number, since it’s soooo likely you’ll want to schedule a reminder call to a phone other than the one you’re using.

It should go like this:

IVR: Hello Shimon. Thank you for cal…
me: wake up calls
IVR: When should we call you?
me: today at twelve forty five
IVR: We’ll call you on Tuesday, November 21st at 12:45pm Eastern Standard Time. Hang up now to finish, or say “timezone” or “start over” to make changes. You can also say “note” to add a voice note, or “recurring” to make this a recurring reminder.
me: <click>

Blending prompts together, and allowing revision rather than requiring confirmation, reduces interaction time in half. Blend this with a simple web interface and you’ve got a self-contained, obvious product that I’d wager people wouldn’t mind paying $5/mo for.

Why not just use your cell phone’s built-in alarm or calendar? I think this would be significantly faster to use, and it would have a consistent interface across different phones. But maybe that’s not enough to make this an idea worth pursuing. I’m not sure.

→ Read the book in PDF format

When I took Physics AP in high school, I made a study guide. As we moved through the syllabus, I’d read a book chapter and review my notes from class, and add to my study guide before each test. At the end of the class, I had a 115-page document that was quite helpful in preparing for the AP exam.

The book is very terse and features lots of equations but only two diagrams. It focuses on presenting and logically connecting the most salient concepts and equations in the Physics AP C coursework, but isn’t intended to be a primary textbook. It also contains some weak jokes; I apologize for any bad writing by 17-year-old self.

I intended from the start to publish the book online, under a Free license. But in the 7 years since I took the exam, it I’ve never gotten around to it. Until now.

Presenting Physics AP in Review:

• Physics AP in Review, version 1.0, PDF (115 pages; 415K)
• Version 1.0 source code for LyX/LaTeX and .eps diagrams (98K)

The book could probably use some tidying up or, if it’s close to useful, a new maintainer who still has an interest in AP Physics. I hope it’s useful to someone!

In a conversation with the Count, I learned a new word:

habernasher (n): one who is in the habit of eating men’s clothing.

Don’t ask. But feel free to suggest your own new made up words!

Over on Reddit, there’s been lots of buzz about Erlang recently. Yet Joel Spolsky didn’t even mention it in a post today about languages for enterprise web apps. This must be because it’s so thoroughly proven, respected, and well-established that you should use it for everything.

After all, I think there are some Swedish phone companies using it for some phone-related apps. And phone-related apps involve zillions of messages per second, so it will definitely be scalable.

Plus I saw this chart where YAWS is red and Apache is green and blue, and red does way better than green and blue. I’m not sure how the test methodology relates to anything you’d actually see in real life, but at least there is quantified evidence that erlang does better than apache at something.

See, the important thing is that you know it’s trustworthy because they made the critical parts so memorable, rather than concentrating on the complex and confusing methodology. After all, what indicates enterprise-readiness better than the existence of an executive summary?

P.S. Someone told me 37signals is developing their next webapp in Erlang. Pass it on!

Josh and I have shut down Frassle, a blogging system we built in 2003 and 2004. Now at frassle.net there is a static mirror of all blogs having 7 or more posts. (A new frassle blog came with 5 posts explaining the system; if you made two additional posts, you qualified for inclusion.) We also set up some redirects so that URLs registered in search engines or linked from other sites should still work reasonably.

I’ve imported my posts into this new blog, and will continue blogging here. Welcome!

A couple of weeks ago I bought myself a little stovetop espresso maker. Since then, I’ve acquired a coffee grinder and some beans that don’t suck, and have been drinking pretty good, wicked strong coffee every morning. So this morning, like many before it, I went downstairs, cleaned up the various steel components, added water, added coffee, screwed the pieces together, and set it on the stove.

Ahh, I thought. With this latest adjustment to the grind setting and the correct water level in the pot, I will be getting the best coffee possible from these beans. I tasted a bean… hmm, not bad, but something a teensy bit darker and possibly more fresh might reduce the bitterness of the final result. I paused to reflect on my sophistication in conjecturing subtle relationships between inputs and outputs in a process executed daily by millions of Italian households.

The stovetop espresso maker is a cool little device. Water goes in the bottom. A funnel with a filter at the top holds the coffee grounds. As the water in the bottom chamber is heated by the stove, steam is produced. The steam, having much lower density than liquid water, creates pressure in the bottom chamber that drives the coffee up the funnel’s narrow end and then through the coffee. Then another funnel in the upper chamber leads the coffee into a holding area where it awaits trial by special coffee tribunal. Er, I mean, where you can pour it into a cup and drink it.

The whole brewing process takes about 5 minutes. Most of the time is spent waiting for the water to heat up; it only takes about a minute to gurgle through the coffee once it gets started. And here I struck on the finest, most sophisticated revelation of them all: now that I’m a certified stovetop espresso guru, I don’t need to stay in the kitchen, waiting for the first signs of incredibly dark liquid pouring to the top chamber. I could go do something else, and as long as I came back within about 3 minutes, I’d be just in time to let the brewing finish and pour myself a delicious cup of lighter fluid for my mind. So I went upstairs to read “just a couple of things” on my computer.

After about 30 minutes, I became aware of a faint burnt-coffee smell. Whoops. Perhaps I’ll be less absent-minded after I get my morning caffeine.

link

BarCamp Boston is this weekend at Monster Worldwide in Maynard. Over 150 people have registered to participate, and I'm sure it's going to be stimulating and fun. I've been so busy working on it as an organizer that I've hardly had a chance to write about it… but keep an eye on Geeks in Boston for updates.

link

You may have seen the MoveOn petition for Congress to preserve network neutrality. The major broadband internet providers — Verizon, Comcast, and AT&T (SBC) — have been lobbying Congress for something. What have they been lobbying for? Abolish "network neutrality"? Write a law so they can do something legal that they already do?

There's a good chance you get your home broadband internet connection through a cable line (especially if your provider is Comcast, Cox, Time Warner, RCN, or Adelphia). You may also get cable TV through the same wire into your house. The cable TV is quite possibly digital. It is sent over the same wire into your home but frequency-decoded by your cable box. Comcast pushes regular internet bits through one frequency and special "cable TV" bits through another. Are the big evil companies asking for something different?

I think it's quite just for Comcast to sell two kinds of service on its network in this case. Are they, Verizon, and AT&T asking for a power they don't already have? What are their motives? What are the specific implications? The debate on "network neutrality" is woefully underinformed on any of these points.

I don't think any of these companies are trying to be nice. They are huge corporations that earn slimming margins providing a commodity service. I have no doubt that they would love to work out some deal with the government that guarantees them a comfy profit margin. But are they doing that here?

If you have any clue, let me know. This whole debate is utterly confusing.

[Thanks to my coworker Todd Underwood for giving me the digital cable TV example.]

A few months ago I gave up working on Frassle, the experimental blogging platform that hosts my blog and a few others. Frassle was a fun and rewarding project: it put me in touch with the very vibrant world of bloggers and social software development, it gave me a reason to present at OSCOM in Switzerland, and it even earned a few passionate users. But I ultimately decided to cease working on frassle. A couple of people have recently expressed interest in picking up where I left off, and that's given me reason to reflect on what I've learned and where it might lead next.

The main reason I stopped working on frassle is that I started putting my time into other projects like voo2do (an easy-to-use online task management application, much more popular than frassle ever was) and my day job. Frassle also has some performance problems: the database structure has too many triggers and a seriously slow full-text search system. The frassle studio—which I think is a good idea that still hasn't been done well—turned out to be perform incredibly badly when implemented on top of a relational database, but could perform wonderfully on a custom sort of publish-subscribe DB. These issues left me in a spot where I felt I'd need to do a lot of rebuilding before I could add nifty, user-visible features.

Yet overall, the core problem with frassle is that it tries to do too many things. It makes conceptual sense to integrate blog publishing, aggregators, republishing, and semantic correlation. But it doesn't make practical sense. In trying to tackle all of these issues at once, I was never able to be the best at any of them. That weakness almost completely prevented users who tried frassle from ever coming back, because when a user is trying a web app, they're not looking for long-term conceptual potential. They're looking for tools that solve a specific problem or make life noticeably better in a matter of minutes. If an app doesn't walk a user through a pleasant experience in the first 5 minutes, that user will not come back. And if you can complete the experience in less than 5 minutes, you will roughly double user retention for each minute saved. I didn't fully appreciate this adage until voo2do, which took a lot less novel thinking and implementation time than frassle, got more users in a week than frassle had in two years.

So what would I do differently if I could do it again? Probably one of the following three mutually exclusive goals:

1. Focus on the inter-blog category mapping feature set only. Create a form where you can paste a blog URL and get content from other blogs that is relevant to the given blog, or specifically relevant to categories within that blog. Use the algorithms from
http://frassle.net/help/welcome and a huge database of RSS feeds with a custom aggregator/analysis layer; a relational database would even work OK.

2. Build a better Frassle studio. Everyone produces RSS now; sites like Pageflakes and Netvibes and Google Customized Homepage are a dime a dozen; but none of these sites let you weave content from multiple sources into a new community website. At some point the people making these tools are going to realize that they're awfully close to fulfilling a non-solipsistic need, and Google Customized Community Website will be born (or bought). Just mix together the tiles-of-content model with the ability to create custom message boards, and you'll have a community website contruction system that offers more power than the average blog, but isn't much harder to user.

3. Social intranet in a box. Now let me partially retract my earlier advice. The one market that *would* be well served by an integrated suite of blogging/sharing tools, even if they weren't all best-in-class apps, is the intranet market. There are tons of LAN-connected collaborative organizations whose members don't want to communicate on the public internet. If you can give these people a bootable CD that they can pop into a spare PC, turn on, and instantly start using as a collaborative intranet server, they would line up around the block to worship you. Nobody gives much thought to usability in intranet applications because the sales cycle is slow, expensive, and managed by IT departments who never use the software and thus don't care if it's excruciating. But if you made bootable CDs that tech-savvy non-IT folks could pop into a spare PC, you could seed a revolution in intranet apps that don't suck.

Sometimes I feel like frassle was my big chance. It had some of the ideas of tagging and social bookmarking even before del.icio.us, and if I'd thought to focus on those parts, maybe I'd have a top 500 site, a yacht from Yahoo!, and 15 million blog groupies. But if frassle fell short because my practical skill didn't match my creative vision, I can only hope that I've learned enough to be get things done without sacrificing much of that vision. I'll just have to work and see.