Visit Your Local PBS Station PBS Home PBS Home Programs A-Z TV Schedules Watch Video Donate Shop PBS Search PBS
I, Cringely - The Survival of the Nerdiest with Robert X. Cringely
Search I,Cringely:

The Pulpit
The Pulpit

<< [ Inflection Point ]   |  The Five Percent Solution  |   [ The Eyes Have It ] >>

Weekly Column

The Five Percent Solution: Who needs another chat client? You do.

Status: [OPEN] comments (20) | add a comment
By Robert X. Cringely
bob@cringely.com

The Pulpit Poll

Have you ever used Mailinator or Talkinator?

Yes: Why wouldn't I?
No: Why would I?

Skip this one and see results

This column is about a new chat system called Talkinator, which I find very exciting, but to do it justice first I have to cover some of the emerging -- but not often recognized -- realities of Web 2.0 that make a Talkinator even possible.

Bill Gates used to worry about Microsoft losing its monopoly overnight because of a technical mistake. We all laughed. We laughed because Microsoft had such financial and sales clout and had the executive suite of nearly every customer company so snowed that they seemed unassailable. But on some level Gates was correct and we've seen that proved by Google.

We didn't need Google, or didn't think we did before Google came along. I don't recall sitting around complaining about Alta Vista and Excite and the other pre-Google search engines, which seemed to do a pretty good job in their day. But then Google came along and was clearly better -- enough better that we all jumped.

How much better did Google have to be than Alta Vista to replace it in the minds and mice of most users? I argue five percent better is good enough. In a market where products are presented as services and those services are ad supported and don't cost users any cash, there is almost no exit barrier. The system has no friction, no stiction. Five percent better is enough to steal that kind of promiscuous market. And five percent isn't much -- a little better UI or server or just a slightly different idea can be enough.

Web 2.0 made this trend even more pervasive, because now applications could be built of other applications, many of them open source. Getting five percent better could mean an idea realized through a mashup with almost no real work, or that was what we have told ourselves. But the reality is that for true success real work is still required, and that's one of the Web 2.0 white lies that need to be exposed.

Web 2.0 makes it easier to do things, but not easier to do them well. You need a good idea, good building materials, and most especially a good carpenter to put it all together. I sense that the Web 2.0 market is maturing and more good carpenters are starting to appear. One of those carpenters is Paul Tyma, the author of Talkinator.

I have written about Paul before in his role as author of Mailinator, another novel web service. That column is among this week's links. Mailinator is an e-mail system that requires no sign-up. The idea is brilliant: if you don't want people to know your real e-mail address, just make up a mailinator address (stinkyface@mailinator.com, for example). Mailinator addresses are useful to give when you don't want to be data-mined and bombarded with offers as the world discovers you are interested in pre-Columbian art or original pictures of Betty Page. Anyone can check messages for stinkyface, but most Mailinator mail is never read because it is spam. The business model here is both simple and modest: show ads to the one percent of Mailinator users who actually DO check their mail.

In order to make money with this business model you need a LOT of traffic, which Mailinator fortunately gets. According to Paul, the Mailinator server receives about 12-15 million messages or 28 gigabytes of mail per day, 99 percent of which is never read. The one percent that IS read means there are at any moment about 150 active users on mailinator.com. The volume of mail coming in has hit as high as 2,000 messages per second.

There is an important lesson in Web 2.0 economics here. Mailinator runs on ONE server. That server is in a rack at Serverbeach and would cost under $100 per month if Paul actually paid for it. But by running a link for Serverbeach on the Mailinator page, Paul gets free service whenever one of his users becomes a Serverbeach customer through that link. His traffic volume is so high that the referrals mean he will never pay a cent for that Mailinator server. So the server is free, the traffic volume is HUGE, and even with that one percent duty cycle the site makes good money from AdSense ads alone.

So why aren't there more Mailinator competitors? There are plenty, but they come and go. The reason they come is because the idea is clever and easy to implement: rent a server and run Sendmail and some scripts. The reason they go is because you can't run Sendmail and some scripts on a single server processing 2,000 messages per second while 150 people read their mail at the same time. Mailinator is a nice little business if you can run it on a single server, but you can't support enough users on a single server if the service is built as a mashup. So Mailinator, which has been rewritten now three times, is 100 percent custom code, highly optimized for what it does, which is the other lie about Web 2.0: mashups often don't scale well.

Sendmail is an Internet bogeyman. Mail is so hard to do well, we're told, and Sendmail is so good (and open source to boot) that hardly anyone ever goes beyond it. We'll put a front end on Sendmail, maybe a webmail interface, but the server, itself, generally goes untouched. With Mailinator Paul Tyma did the unthinkable and threw Sendmail away, replacing it with a truly modern mail engine that someone smart ought to buy.

Which brings us to Talkinator, Paul's new chat system. This is quite specifically text chat and involves no voice or video, just letters and words. And like Mailinator, Talkinator is a no-log-in system.

Why would he do that? Why have no log-in? The better question might be why do the other chat systems HAVE log-ins? They want you to register so those systems can track you and make money from your chatting habits. They have log-ins so those sites can be more useful, maybe, but mainly so they can more effectively USE YOU.

Talkinator is different. Imagine going to a website is like entering an unfamiliar house. Web surfing is, for the most part, a solitary experience, even though there can be hundreds or thousands of people on the same page at the same time. With Talkinator you enter the room and ask, "Is anybody home?" Wow! There are 138 people there right at that moment and with no registration you can call yourself anything you like and start chatting immediately.

Embed Talkinator in YOUR web site and your users can communicate in real time with each other and with you. It's free.

Now imagine you run an Elvis tribute page and a bunch of your friends run Elvis tribute pages, too. If you all embed Talkinator in your sites and choose the same name for your chat room YOUR USERS AND YOUR FRIENDS' USERS WILL ALL BE IN THE SAME ROOM.

But what if the linked sites or their users are from different countries and different cultures? Talkinator has a bot that leverages Google Translate so users with different languages can still communicate.

I use Google Translate to communicate with my cleaning lady, who speaks far better Portuguese than I do. But Google Translate doesn't always tell her exactly what I mean, so I began doing recursive translations where I would translate my instructions into Portuguese then back into English so I could fine-tune the text until I knew it was perfect. As a favor to me Paul implemented this in Talkinator as what he calls a round-trip translation option so misunderstandings are minimized.

Arguments are minimized, too, since Talkinator simply won't allow swearing. Try it.

So who needs another chat client? We all do. Talkinator isn't intended to replace ICQ or AIM or any other chat system with zillions of users. It is intended to help people with common interests meet each other and communicate on the Web. THERE IS NO BUSINESS MODEL. Talkinator is an experiment, but it is so useful that traffic will build to a point where, like Mailinator, a single author can make a fair living just by having fun.

Getting back to the Bill Gates market dominance model, Talkinator has to be a huge headache for companies doing something similar. Look at Meebo, which presents itself as the web intersection of AIM, Yahoo Chat, Google Talk, and MSN Chat with some ads thrown in. Talkinator was written by one very good programmer, runs on one free server supporting thousands of simultaneous users and can be scaled to any number of users with additional free servers. Talkinator could be turned into a Meebo competitor over a weekend.

That puts Meebo's recent $200 million valuation in a different light don't you think?

Comments from the Tribe

Status: [OPEN] read all comments (20) | add a comment

What? No swearing? How is that going to be successful?

Partners in Grime | Jul 29, 2008 | 9:55AM

Mailinator is pretty a pretty neat toy, but if you read the architecture doc that the engineer wrote, you'll see that it's not a fully implemented SMTP server. It cannot send mail. Worse, it drops attachments/images, etc. Finally, it stores all email in memory and deletes old messages. Now, I'm mighty impressed with this java hack, but it's not a high-performance, general purpose SMTP engine. Sendmail, postfix and Qmail are all beasts, but they are full of features that serious organizations need. Thanks for highlighting this developer's work.

Joe Johnston | Aug 01, 2008 | 5:16PM

I think it's fairly easy for people argue without swearing. It'd be nice if you could block IP addresses.

If I were to put this on my website, people could either spam or troll my potential customers and current users.

So now Captchas and their problems.

Alex Birch | Aug 01, 2008 | 6:12PM
[OPEN] read all comments (20)

ADD A COMMENT

Ground rules for posting comments...

  1. No profanity or personal attacks, please.
  2. Please restrict your comments to the subject of the column and directly relevant topics.
  3. Be more funny.
  4. Those who violate these ground rules will have their comments deposted (which is a bit like being deported, only you don't have to leave the country).

name:

e-mail:

NOTE: Your email address is for internal purposes only and will not be published, shared or sold to other entities.

url (optional):

Comment (br and p tags are not necessary for line breaks)