« January 2007 | Main | March 2007 »

February 2007 Archives

February 4, 2007


Where have I been... I'll tell you where I've been. I've been looking at dead bodies in Las Vegas and it SPOOKED ME OUT.

I missed Bodies...The Exhibition when it was in San Francisco a while back. Last week we were at the NAA convention in Las Vegas, and Burns tells me the bodies exhibition is at the Tropicana. I had to go. I'd read about the crazy process where they take cadavers and fill the veins and stuff with plastic and turn the bodies into life-sized, real models of themselves or something. So Burns and I take a hike up there from Mandalay Bay one day before our show.

Tolles wouldn't come, he was too creeped out.

The exhibition was amazing, even better than I expected.

Imagine a person's circulatory system, filled with plastic, holding its shape, with the the rest of the person dissolved away so only the red-colored blood vessels remain, hanging 3D in space.

The fetus room had its own special warning, so you could go around and skip that part if you wanted to. Rightly so. Yuugh.

The respiratory system room had a big plexiglass box with a hole in the top of it. The box was filled with hundreds of packs of cigarettes. Next to the box was a case with two preserved lungs, a healthy lung and a black-tarred smoker's lung. A sign over the plexiglass box said "Quit now." While Burns and I watched a woman dropped her pack into the box. I wonder if she'll really quit. It's hard.

Spending an hour looking at partially-dissected cadavers really started to haunt me about my own mortality. I was wondering who all these exhibited cadavers belonged too. I mean who were these people that their bodies ended up in a traveling show in Las Vegas with a bunch of people gawking at them. This isn't just Hamlet holding his buddy's skull, it's surreal on a modern scale.

The last time I felt this way was in Paris, when I visited the catacombs. Apparently in the 18th century the dead of Paris were overflowing the cemetaries, so monks dug up all the bodies, cleaned off the bones, and neatly stacked them in various ..patterns.. against the walls in a network of tunnels buried beneath the city. You can take a tour. (Take a close look at that picture. Those skulls are arranged in the shape of a heart. What were the monks thinking?)

The place is genuinely creepy, all wet-dripping stuff from the ceiling and skulls and plaques with ominous messages of doom every few paces.

While I was there a girl from a high school tour group freaked out and ran past me screaming "I have to get out of here!" She was going the wrong way.

That was ten years ago. I think it would be even more spooky to me now. You don't see any old people touring the catacombs.

February 6, 2007

98% spam

Absolutely horrifying email spam stats from our new vp ops at topix.

We receive 25,000 mail connections per day; each connection is an attempt by some machine to send us mail. Of those 25,000 connections we are able to reject nearly 80% of them outright since the IP address of the originating machine is registered as a known spam offender. Using various additional checks and methods we are able to reject about 98% of all incoming connections before they even have a chance to send us message content. Of the 2% that do make it to the point where they send us a complete mail message we are able to reject 25% of those as containing spam, a virus, or an unsafe attachment type. That means that in the end only 1.5% of all attempted mail connections actually result in delivered mail.

That's push email. Consider the day in our near future when 98% of the http fetchable web is spam. Auto-generated text, on-the-fly scraper-reconstituters, and so forth.

The bright side: web spam is an evolutionary force that pushes relevance innovations such as trustrank forward. Spam created the market opportunity for Google, when Altavista succumbed in 97-98. Search startups should be praying to the spam gods for a second opportunity. :-)

February 7, 2007

Topix vs. the newsroom model: enabling mass many-to-many communication

Getting hyperlocal community started on the web is hard. It's tough to do for a single community. This is where successes like Baristanet and Mike Orren's Texasgigs stand out. You have to start with some seed content to draw an initial audience, and then get them talking back to you, and talking to each other. It's a tricky boot-up process to manage, and many attempts have failed.

It's even harder if you're trying to take on thousands of local communities all at once, with a single brand.

Local newspapers by all rights should own their local communities online. But there are 1,500 local newspapers in the US, and most of them don't have a lot of spare resources to experiment online. Usually most of their effort is focused on the product that keeps them employed -- print. And I'll say it -- culturally they don't really like the web. Newsroom-driven attempts to deal with the web are usually one-way and read-only. Of course. Go to newspaper industry conferences and you'll find they're still talking about nonsense like electronic readers to view images of the print newspaper, and charging money for online subscriptions. And when it comes to facilitating conversation with the community, newsrooms tend to get seriously uncomfortable with the realities of dealing with the net public. They typically pull community the minute things get hot -- which, ironically, is when you need discussion and debate the most, and when the discussions often become the most interesting.

There's an underlying assumption in common between the newspaper's online efforts and the single-community hyperlocal startups: that there will be many local brands online. Currently there are over 3,000 local news brands in the US, if you include both newspapers and local TV stations with newscasts. That made sense when geography defined the radius of distribution around TV towers or newspaper delivery trucks. But put your content online, and that radius goes away. Distribution expands to the edges of the net.

Are 3,000 community websites sustainable? Will online consumers really accept 3,000 local brands? Or will a McDonald's model win instead, subsuming the thousands of mom & pop hamburger stands?

Topix is betting that the national play will ultimately win.

Single-community hyperlocal sites tend to have passionate, involved editors and participants. But the labor and overhead costs are high. And they don't generally scale -- to tens, hundreds, or thousands of additional communities. So the costs of the overhead have to be borne by the revenue potential of just one or a small number of communities.

The challenge for a national hyperlocal play, on the other hand, is to get quality in the long tail of geography. Can that be done from an office in Palo Alto? Won't local community development suffer without an editorial office based in every city and town?

But this apparent deficit may in fact be a strength.

Single-community hyperlocal sites are still rooted in the model of the print newsroom. This is a one-to-many mindset that empowers a few to determine what the many will read. It made sense when the distribution network started with an expensive printing press.

But the Internet is not about one-to-many. It's not a printing press or a TV tower. The Internet is the first mass many-to-many communication medium. Users aren't on their community site just to hear one person talk. You can be the greatest local journalist in the world, but reading your output takes 5 minutes a day. Users are there to talk with each other. To learn from, gossip with, and argue with their neighbors. Users provide their own draw. The necessary job of the 'editor' here isn't to run the conversation, like a teacher in a 6th-grade class. It's simply to make sure that the conversation 1) gets started, and 2) doesn't completely run off the rails.

Scaling this to enable millions of people to talk with each other in thousands of simultaneous parallel conversations is not simply a matter of putting up lots of empty message boards and hoping it all works. You need a lot of seed traffic, as well as topically- and geographically-segmented content to prime the pump. You also need a social architecture that draws visitors into discussion with each other. Challenge #1 is getting the party started.

Challenge #2 is keeping it safe and on-track. A free concert in the park with 30,000 music-lovers is great, until the lights go out and the police aren't there. Then it can get ugly really fast. Not because most of the participants are bad people. But because a few bad apples can ruin things for the majority.

This has happened to nearly every online community system, from Usenet to the comment threads that the Washington Post took down, to the news message boards that Yahoo recently took down.

We've spent considerable effort at Topix on an architecture to not only get online communities booted up, but to let them socially scale. Effectively dealing with spam, hate speech, profanity and trolls isn't just about maintaining the quality of the commentary. Scalable moderation is essential for enabling growth to larger and larger audiences. If you don't keep the quality sufficiently high, you stop adding users.

We do a lot of this at the software level, with layers and layers of filters to catch all sorts of bad stuff. We also have centralized human editorial overseeing the whole show. The simple idea: get rid of the bottom 5% of posts every day.

Starting from zero a little over a year ago, we now have over 1,000 active forums (an active forum being defined as one which receives at least 5 posts per day). There are 30,000 local communities in the US, so by one measure we've hardly made a dent in nationwide hyperlocal. But 1000 active communities is no small achievement, when you consider how hard it is to start just one. Local community now represents 35% of Topix's traffic, and continues to grow double-digits every month.

Community-driven content has been so successful for us, in fact, that were going to be re-organizing our entire site to focus on it. I'll have more to post about this as we get closer to the date of our big relaunch. :-)

February 10, 2007

Boost RF range with your head

Here's a weird tip I learned way back from a hardcore Sun engineer, Ben Stoltz... You can use your head as an antenna to boost the range for little RF devices like car key fobs, garage door openers, etc. It sounds crazy and I didn't believe it until I tried it myself. Stick the device under your chin and hit the key... I can double the range on my car RF key this way. It really helps to find your car in a garage when you've forgotten where you parked. Or to hit the garage door signal when you're slightly out of range.

Of course you have to not worry about what this is doing to your head. Is it just something about the shape of your skull, or is it the quasi-electrical circuitry of your neural mass that is serving as the antenna to amplify the signal? I don't know. Jeezus. I tried to swear it off but then I was looking for my car once, and it just works so well... Don't think about it, I tell myself.

I tried to use this trick to get my blackberry to sync to the ground from the 6 hour flight I was on today, but no luck. I would get partial signal for a few seconds but then it would fade out. I guess if you're going 500 miles an hour it's not just the 35,000 feet keeping you from hitting the towers. Where's our in-air wifi?


February 12, 2007

Facial Action Coding System

I mentioned before reading Blink and becoming fascinated with studying facial emotion with the Facial Action Coding System (FACS). I'd played with some of the online tools like Artnatomy, but apparently full FACS training takes 80 hours and requires a bunch of video; you can't learn it from a book since you have to be trained to recognize fleeting subtle expressions and what they mean.

So I ordered a training CD from the lab of Paul Ekman, who is one of the researchers who developed FACS and it finally came.

Micro Expression Training Tool

While most facial expressions last for two or three seconds, micro expressions last a fraction of that -- 1/25th of a second. These are signs of emotions just emerging; emotions expressed before the person displaying them knows what he or she is feeling, or emotions the person is trying to conceal. You can learn to spot these micro expressions and have access to this valuable information.

Subtle Expression Training Tool

With SETT -- in under an hour -- you can train yourself to see very small facial movements that often appear in just one region of the face: the brows, eyelids, cheeks, nose or lips. These small movements may occur when an emotion begins gradually, when emotions are repressed or when a person is deliberately trying to eliminate any sign of how he or she is feeling, but a trace still remains.

Understanding the code-language of the face seems like a great way to improve communication, not to mention being able to spot lies, false smiles, contempt, and the like. This seems like it would be useful in business, relationships, all sorts of situations.

I just started working through the exercises on the CD today. We'll see how it goes. Unfortunately this CD isn't full FACS though, I may need to hunt around for additional training materials.

February 13, 2007

The Failure of We (the) Media

In the wake of the latest We Media event last week the usual round of self-flagellation by a group of attendees is occurring.

David Cohn wishes We Media was an unconference. Scott Karp acknowledges that media companies need to make money, but bizarrely refers to that as an "idealogical agenda". A BBC exec calls it groundhog day. The staid Mark Glaser even throws a few rocks (one at me!), with a big pile-on in his comments to boot.

Andrew Nachison and Dale Peskin put on a solid show. They make a real effort to have the panels interact with the audience. This is not easy -- you get 200 people in a room for 90 minutes, how much interactivity do you really expect? But they managed to pull it off and the results were far more successful than many other panels I've been on. They also get really amazing attendees. Major publishers, tons of senior media execs, a dozen startup CEOs, VCs, Pulitzer-winning journalists, and an eclectic spectrum of media spinners like documentary filmmakers, artists, social activists, and the like to keep the cocktail mixer from getting too businessy and boring. If you can't find someone interesting to talk to at this thing, it's not the event's fault.

So what's the problem?

The problem is that the hopes that Dan Gillmor raised for the media industry in his book -- which kicked off this whole business -- have largely failed.

Tremendous excitement followed the publishing of Dan's We the Media (the conference's namesake). It accompanied the trumpeting of a new model of media by the newsy press, and the rise of blogs with attendant breathless hype.

Unfortunately, after doing the author's victory tour, Dan then attempted to put his ideas into practice in a business venture. I suppose there is some due credit for having the courage to cross the line from a long career as a newspaper journalist (observer) to become a startup founder (participant), and try to prove the viability of his alt.media business plan outlined in the book.

But, like nearly every News 2.0 venture so far, Dan's Bayosphere was a failure.

He has a lot of company. The dog's breakfast of new media startups includes Gather, Backfence, Newstrust, Daylife, TailRank, Associated Content, Pegasus News, Tinfinger, Findory, Inform, Newsvine, Memeorandum, NowPublic. The highest distinction on this list is to be one of the few still spoken of in the present tense (or present perfect -- "They haven't yet succeeded...")

And yes, I would include Topix here as well. We are, in fact, the most successful News 2.0 company, with over a million pageviews/day, 10M server/4.6M Comscore uniques, a million participants in our forums, a $60M exit, yada yada. But, we can face it, even we haven't yet burned down the world, or upended the news industry.

There is actually a media revolution in the works. So what's going on here? By implicit definition, participatory media is non-commercial. If it's commercial, someone owns it, and it's not "we" anymore.

Furthermore, as soon as a new media venture crosses the line and tries to become a business, it either becomes a successful business or a failed one. Businesses aren't about ideology, they're about getting a job done and earning revenue to keep the thing going. Even wild success tends to leave ideology behind. Ideology is the realm of nonprofits and failures.

There is still a power law to success, and the few continue to reap disproportionate rewards, as they always have. Pub media turns out to be a farm league for big media. The bloggers who "make it" look more and more like regular media than "us". They graduate to to the A-list, and start to get lumped in and criticised along with the establishment. Success looks like a sellout to a big media company, or a good business doing job boards and conferences on the side to pay the bills.

Yes, there is a media revolution in the works. But it's messy, it's nasty videos on Youtube, not the neat & tidy civic Welcome Wagon of citizen journalism. You can't quit your job as a journalist and replace your salary with adsense on your blog. You'll be lucky to make beer money, let alone pay COBRA and fund your SEP-IRA.

And big media has been watching, and buying the winning ventures, and building their own platforms to -- yes you're right! -- exploit the new models.

So shut up and keep blogging, or putting your time in as a wage slave in your chosen profession, or keep slugging it out at a startup. But please stop whining that "we" haven't achieved consensus at the latest industry schmoozefest. If you don't know why you're there, you probably shouldn't be. :-)

February 14, 2007

Nothing up my sleeve...

February 18, 2007

Two Cows and Venture Capital

You have two cows. One is male, and one is female. Mike Moritz says he loves both cows and will buy 35% of the pair for $100. After the deal is signed he tells you to kill your female cow, and then says your male cow must produce a baby cow within three months or you're fired. Three months and one day later he fires you, takes your remaining cow, and transfers it into a milking machine company which then goes public on Nasdaq, earning him $10,000,000. Citing a lactation preference in the term sheet, however, he keeps all but $0.10 of the proceeds. "No hard feelings," he says, "and be sure to come back the next time you have cows."
    -- Paul Kedrosky


Creating a new product like the iPod or even the Prius is a far more modest achievement than developing a new process. The former are what we normally think of as inventions, of course. But the latter, at least in Toyota's case, presents a novel way of thinking about work and the capabilities of human organizations.
    -- From 0 to 60 to World Domination, NYT Magazine
"People don't scale: Truly lazy developers let their machines do the work for them... smart developers know that people don't scale-- machines do. If you want it done the same way every time, and with any semblance of reliability, you want the human factor removed as much as is reasonably possible... I ask myself-- how can I make sure I never have to deal with this problem again? If my solution fixes it so nobody ever has to deal with that problem, that's a nice side-effect, too.
    -- Jeff Atwood

February 24, 2007


Markson gnaws at the apparent lack of a hitmaker playbook for online marketing:

Many industries have hit-making down to a science. Hollywood is a good example - from the script to the the cast to the production to the testing to the release to the marketing to the distribution - every step of the way, calculated decisions are made that are designed to maximize a movies success. Same can be said for cheeseburgers, toothpaste, autos, etc. They know what it takes to make a hit and they pursue it. They might not always get it right, but there is some research and science to the choices that are made.

Offhand, I can think of some online equivalents. The gaming industry is a hit-driven business. A game starts with a concept, whether it's Tiger Woods Golf, Lego Star Wars, or some MMORPG Dungeons & Dragons descendant. Get a physics engine, voice actors, motion capture, all that expensive stuff that goes into the multi-million dollar gaming budgets these days. Get it out by July so it it can on the shelves for Xmas. Make sure your distribution agreements are in place so your focus-group tested dodecahedral boxes will be arranged in the proper stacks on the end caps at Fry's. There's a lot of details to get right to drive a gaming hit.

Another: the mini blog empires built by Jason Calacanis and Nick Denton. The model is -- think of a concept for a blog, get a fresh design and a great name, and hire a contract blogger. The pay is $2k/month, you're 1099, not an employee, bring your own laptop, work from home, and if you don't increase traffic 10% month-to-month you're fired and someone else will try... After two or three bloggers have tried, if the site is still not working it gets shut down. Have competent centralized ad sales, launch PR, and a great serving platform. Repeat and scale.

But Mike then talks about distribution strategies:

As far as I can tell there are only three real online hitmaking "strategies": SEM/SEO, viral and syphoning (taking your existing traffic and using promotion on your site to "syphon" it off to a new product.)

I don't think of those as hitmaker strategies, they're only a piece of the puzzle. Those are tactical distribution methods. I'd add pure word-of-mouth as the holy grail here, separate from viral:

  • SEM - purchase traffic profitably via some kind of arbitrage. ex: Shopping.com, NexTag, Monster.com.
  • SEO - rank organically for free search traffic. ex: About.com, Autobytel, Yelp.
  • Viral - product contains a built-in spam-your-friends mechanism. ex: Hotmail (the original!), Friendster, Youtube.
  • Syphoning - brand extension and traffic promotion from a winning business to promote a new product. ex: Gmail. I'm having a hard time with examples here because this strategy basically doesn't work. Zshops, MSN, Live, Endless.com, Alexa, A9, Yahoo 360...?
  • Word of mouth - something so compelling people refer their friends even without a built-in spam mechanism. ex: Google.

Viral and word of mouth often work together -- a spam-your-friends mechanism will only be used if someone likes the service enough to use it themselves, and to recommend it to others. It just lowers the activation threshold.

For most people in our industry this is old hat. Yet still there are routinely sites launched by startups and big companies alike that don't have any of these mechanisms. It's hard to predict success but you can often spot failure in the works when a site launches with zero SEO, no way to sustainably buy traffic, no viral aspect, a ho-hum product, and a billboard on the hill in South San Francisco begging visitors to come. These happen all the time and you wonder what the VCs and founders were thinking. Can web distribution 101 really be unknown to them at this point?

The thing is, the entire startup environment in Silicon Valley and the rest of the net-connected entrepreneurial world is the hitmaker factory. The VCs are the formal, but not exclusive drivers of this show. VCs are the hitmakers. They have time-honed playbooks for how to churn out hits from entrepreneurial ventures in fast-changing markets. Even their apparently-trite maxims actually code for a wealth of wisdom.

Movies and games burn out after a while so the engine needs to keep making fresh ones. For programming problems and online brands, though, once you have a winner you're basically done unless the winner screws it up at some point (e.g. AltaVista -> Google) or the market or technology base moves again.

So once you have Google you don't really need more people trying to make a great search engine; Silicon Valley made a really good one, everyone in the world can use it for free, and they're doing a great job keeping their product in good shape.

That happens all the time. We don't need big dialup ISPs anymore, so UUnet and AOL are in the past. We don't need PC or OS startups anymore so they're in the past. We don't need browser startups anymore since we've all got several browsers that work just fine. Mosaic/Netscape and Spyglass are over. We don't need someone to start retailing shoes or books or X online anymore since you can buy anything you want with a few mouse clicks and have it tomorrow FedEx. Fortunes were made filling all of those needs, but the needs are filled now. Fortunes were also made delivering nutmeg and sugar and fresh lettuce to our kitchens too. But they're done. You have to find a new need to make a new fortune, not solve an already-solved problem.

About February 2007

This page contains all entries posted to Skrentablog in February 2007. They are listed from oldest to newest.

January 2007 is the previous archive.

March 2007 is the next archive.

Many more can be found on the main index page or by looking through the archives.

Powered by
Movable Type 3.33