Philly Emerging Tech – Day 2

Today was the second day of the 2009 Philly Emerging Tech conference.

I was slightly worried about the traffic, so I left nice and early to ensure I’d be there on time today. Fortunately, the weather and trafic were good and I actually knew where I was headed so I arrived with plenty of time to spare. With about 45 minutes until the first talk, I grabbed some fruit and sat down at a large table table to figure out which presentations I was going to attend.

A short while later a towering fella with a moustache came and sat down next to me. I recognized him from the day before. At about 6′ 6″ and the only white guy with a stash, he’s hard to miss. Jeff, it turns out, works for a local company that creates mapping software and he was at the conference to pick up new techniques that he could take back to apply to their processes.

Maybe it was just a coincidence or maybe it was just because he was a giant and I kept noticing him when he was near me, but it seemed like Jeff was never more than 10 feet from me throughout the entire day. It became this awkward running joke because every time I went to the restroom he would somehow wind up being next to me in line. Anyway…

~

Andy Hunt was up first.

When I read books, I highlight the stuff that’s meaningful to me. If, after a few chapters, I haven’t highlighted anything, I usually put the book down and find something else. Occasionally, the opposite is true. There will be so many good lines that I have to stop highlighting, because it defeat’s the purpose. 37Signal’s Getting Real comes to mind.

Andy’s presentation was like that. There were just so many good, practical take aways.

Here’s a sampling of the topics he talked about, which were derived from his recent book Pragmatic Thinking & Learning: Refactoring Your Wetware.

Experts have the ability to recognize patterns in a given context. For example, doctors can sometimes diagnose a patient’s illness from a quick, cursory glance without knowing exactly what led him or her to that specific conclusion.

It used to be a commonly held belief that you could not grow new neurons. You could kill them, but you couldn’t grow more. The problem was that they were basing that conclusion on tests performed on primates held in captivity. If you exposed the same primate to an environment filled with rich sensory inputs, you’d  discover that it was possible to grow new neurons.

The Dreyfus Model of skill acquisition represent’s someone’s journey from notice to expert. An expert’s relationship to the system changes. Experts rely more on an intuition and an internalized body of knowledge instad of rules. They had expert airline pilots document how they made their decisions. When given to the beginners, it improved their performance but when they asked the same experts to follow those rules, it degraded their performance. In the Dreyfus Model, novices are detached observers of the system. Experts, on the other hand, are a part of the system.

If you criticize the system, the expert is more likely to take offense to it.

How do you become an expert? Delibrate practice. Imitate, assimilate, innovate. First you following along, then you make it part of your experience, then you change it. It takes about 10 years to master a subject. He gave Linus Torvald and Motzart as examples. If you imitate a smart person, it actually makes you smarter.

Note: I think Mr Hunt reads Malcom Gladwell :)

Scientists used to thinking that the right half of the brain was used for creative, innovative activities and that the left was primarily used for linear functions. It’s actually a bit more complicated than that.

Linear skills are things that we computer science guys know a lot about. It’s the logical, analytical skills that we use everyday to do our jobs well. Nonlinear skills, however, are the hallmark of the expert. It’s not rational. It aggregates a whole bunch of things together. It’s more about synthesis than analysis.

You process aesthetically pleasing things better.

Saying that your iPod can store 5 gigabytes is a lot different than saying it can store 20K songs. Grandma doesn’t know.

This mad psychologist scientist named Lozonov did an experiment where he had his students relax and listen to hippie music before learning new material. When compared to a control group, they had much better retention.

Did you know… Thomas Edison would take naps with ball bearings in his hand before so that when he finally dozed off they would fall and wake him up. He would then immediately write down on a notepad whatever he was thinking at that instant.

Mediate. It can do amazing things.

People have lots of cognitive biases. We’re not rational beings. Our consciousness gets inputs from our memory, from our sensory inputs, and from our imagination. Our mind is not so good at discerning which though.

The relationships between things are often more important than the things themselves.

Think about the color red and all the sudden you start seeing it everywhere.

When you read, one method to improve comprehension is to follow the SQ3R method: Survey, Question, Read, Recall, Review.

Bobsledders visualize the course before they begin. It helps them perform better when it’s game time.

Multitaskng, as in constantly checking your email, robs you of 20-40% of your productivity.

Context switching, like switching applications on a computer, is a cognitive train wreck.

Organizing applications by the task helps you avoid context switching.

Something about a pigeon banging its head against a tree.

… I thought his presentation was the most interseting out of the 15 or so I saw the last two days. That book is on its way.

~

Next Up: Mike Culver, who gave a great introduction to cloud computing and Amazon’s Simple Storage Service (S3). I didn’t know much about cloud computing beforehand, but Mike made sure to remedy that.

We need more innovative companies like Amazon.

~

After that I attended a talk on Scrum by Jim York. Scrum, from what I gathered, is an agile software development technique that focuses on breaking up objectives into one month groupings called sprints. He had a lot of good lines too:

“To be successful, an organization must produce something of value.”

“How do you create value?”

“Organizations create value when they deliver to the customer:

1) What they want

2) When they want it

3) At a price they’re willing to pay.”

“When task is made king:

1) Management creates organizational structures best suited to accomplish the tasks

2) Efficiciency experts establish the best way to execute the tasks

3) Quality experts implement controls to ensure processes stay within defined limits

4) Workers are trained in processes to complete the tasks

5) Management evaluates performance on process compliance”

“Where is the innovation?”

Management and Quality Assurance, NOT the ones actually doing the work.

Change disrupts process. Most processes are change resistant.

Work is prioritized from the customer’s perspectuve.

Customer is king.

Identify your ultimate customer.

Involve the customer.

~

The third talk was by Jascha Franklin-Hodge, who helped cofound Blue State Digital, the company that led Obama’s tech campaign. Few notes:

Blue State Digital was cofounded in 2004 by four former Howard Dean staffers, one of whom was Jascha.

The company currently has more than 75 employees and 200 clients.

There’s three big parts to what they do: design, which involves incorporating your vision of what you want the users to do; technology, such as the donation process; and strategy, which is tackling how you want to achieve your goals… “Not just taking brochures and putting them online.”

1B emails to 13M addresses

1M SMS

200K offline events planned over the web

35K local volunteer groups

14.5M YouTube minutes for official videos

The Obama campaign raised $770M dollars and more than half a billion of that was online.

He then went over several overarching themes for how they did this:

Drive Action – the Obama homepage was “all about what you can do”. Each medium has its own set of activities. iPhone app would sort your contacts by area code, listing those in key battleground states at the top of the list. 70% of actions come from the top 10% of people.

Be Authentic – emails from people (Barack, Michelle, David Plouffe), not from namesless org accounts. Notice how Google blog posts are by the engineers who developed the product.

Turn Users into Advocates – recognize your leaders and engage them. Invite people to weign in. Collect user input and reflect back on the good stuff. Solicit ideas from the community and actually use them. Connect people with each other.

Be Relevant – Palin’s speech resulted in $11M being donated — #1 contributor. They took advantage of the situation.

Build a Strong, Open Brand – Make it professional. Same logo was used everywhere.

Measure Everything – “If you’re not measuring it, if you’re not testing, you don’t know what you’re missing.”

They did A/B testing on donation forms to see what people responded to the most.

Not a lot of tech savy people in the political world

~

After that I attended one more Scrum presentation, this time presented by Mike Vizdos. For some reason, I couldn’t help but think of Crum from Ahh Real Monsters whenever anybody said Scrum. Come on… “Scrum Master”? IMHO, they should have picked a better name.

~

And last, but certainly not least, was Andy Hunt again. Few more noncoherent notes:

“Without excellent personnel, even good to excellent proceeses can only achieve marginal results” – Casper Jones

Talent matters. People make a big difference.

The real world is messy.

Kaizen – Feedback and continuous improvement. Have a short feedback gap.

Systems Thinking – Problems with any person on the team are problems for the entire team; it’s all one system.

Risk management – minimilism is a good approach for dealing with risk.

ROI – make your product useful sonner rather than later.

Continuousness – Don’t take too big of a bite at any one point in time.

((((do something!) small) useful) now!) – Bob Berner, father of ASCII

Feedback – “Agile development uses feedback to make constant adjustments in a highly collaborative environment.” – Practices of an Agile Developer

Ask yourself periodically:

1) Why am I doing this?

2) Does it have to be done this way?

3) Does it have to be done at all?

~

On a final note, as I was leaving the parking garage there was Jeff again.

He told me to get my belts replaced. =)

Philly Emerging Tech – Day 1

Today was the first day of the annual Philly Emerging Tech conference. Here’s a quick rundown:

I left about an hour and a half early to allow plenty of time for Philly traffic but despite my GPS’s estimate of a 40 minute commute, I wound up arriving 15 minutes late. The bad weather and a wrong turn into Camden didn’t help anything either.

Today’s keynote speaker was Michael Tiermann, Vice President of Open Source Affairs at Red Hat and President of the Open Source Initiative. To give you an idea of his vision, here are a few quotes from his presentation:

“Lots of innovation results in an increase in productivity”

“Now that we can do anything, what should we do?”, quoting Bruce Mau

“How can we be better?” quoting JP Sloan

“Leave your system open to innovation”

Did you know… Somebody did a study of contributions to Apache and calculated that 1 developer did about 20% of the work, 5 did about 50%, 15 about 80%, and an amazing 388 people to do all 100%? Also, proprietary software averages 20-30 defects/1000 lines of code. Open source: less than 1. The linux kernel is about 5M lines of code. An automated software scan came up with 985 errors and with the help of the community, they were all fixed within six months. Now compare that to Vista, which is estimated to be about 50M lines of code, which does not have an extensive community to help fix what must be at least a few hundred thousand lines of defective code.

He said something else that I thought was good. I forgot the context, but it was something like “The cost to the developer is less than the value to the customer.”

When he finished I went to an introductory presentation about iPhone software development by Bill Dudney. I never really appreciated how easy it is to create an application’s interface. I thought you had to program the behavior of the tables, the sliding buttons, etc. Turns out most are just customizable controls. He walked a packed room through the creation of a simple app in under 40 minutes. He was very well spoken and definitely knew his stuff.

I bounced around a bit during the next hour. I started off in a presentation about Android development then went to Exhibitionism in Software Development and finally wound up in a talk being given on the importance of accessibility in web development.

After that was lunch. I thought we would have to leave to go get lunch, so when I came out of the accessibility talk and a buffet was already set up, it was a pleasant surprise.

I ate with a few other people from the Philly on Rails meetups–Chris, Jon, Angel, and Randy. Colin, Alex, Aaron, JP, and a few others were around too. Also met Chris, the CTO of a Philly startup called Vuzit that has created a novel Ajax-based document viewer.

After lunch was a talk called Innovation in Ruby given by Jason Seifer and Gregg Pollack of Rails Envy fame. Their presentation was excellent both in terms of content as well as how they spoke and interacted with each other. For some reason I kept thinking “Batman and Robin” the whole time. Anyway, a lot of it was over my head, but I left with a much greater appreciation for the brilliant work being done in the Ruby and the Rails communities. I also briefly met Ezra Zygmuntowicz, who apparently founded Engine Yard and created merb. Nice guy.

Next was John Resig of jQuery glory. Before the talk I asked the guys why use jQuery over Prototype. I don’t remember what Randy said, but it was something poetic about how code just flows from his hands or something to that effect. John’s talk was good, despite the Public Address problems that resulted in us hearing the presentation being given in another room and eventually a full blown rock song. Next project is going to be with jQuery. I’m convinced that it kicks ass.

Last but not least was Mike Culver from Amazon Web Services who spoke about and demoed Mechanical Turk. I thought his presentation was the most interesting one all day. What an amazing technology.

Tomorrow: Day 2, where I continue to learn more about just how much I don’t know. :)

Back to Work

Somehow about two weeks went by without a update to Domain Pigeon. Domains were still being released (that’s all on autopilot now) but I hadn’t made any changes to the site. I didn’t think it was that long, but when I started working on some new features tonight I realized just how long it had been since I last opened TextMate or ran a script/console command.

As is normally the case, the time off resulted in some innovative new ideas for the site. With those in mind, I’m refreshed and ready to start working again.

On an unrelated note (maybe), I came across an excellent Washington Post piece titled ‘Pearls Before Breakfast‘ which is much needed reminder to slow down and enjoy to the ride.

Updates

Not much to report on.

I spent a few hours this week at a local sandwich shop reading Beginning Python and my initial impression is extremely positive. My next project will likely be made with Python and Django.

I’m also working on some potentially big updates for Domain Pigeon, which I hope to roll out in a week or two. I realized that I’m in a great spot to just experiment with the site and see what happens. If the changes work, great, if not, no big deal. After all, failures are excellent opportunities to learn what not to do which can often be just as valuable than knowing to do. :)

A Brief HackerNews Traffic Analysis

On Monday morning I added a link on HackerNews to a recent post on this blog. Happily, a short while later it was on the front page, where it sat in the top 30 for about a day and a half until it dropped off. Using Google Analytics, its possible to do a little analysis of what kind of traffic you can reasonably expect from a popular HN submission. This blog gets only a handful of visitors on a normal day, so the results sure be mostly pure.

It’s important to note upfront that it’s hard to draw too many conclusions from the data. What time you post, what day of the week, the provocativeness of the link title, how high it gets in the rankings, how long it sits there, the comment to point ratio, your nick’s reputation, and a lot more could drastically change how many people click the link and make it to your site.

Without further adieu, here are a few traffic stats for this blog from Monday and Tuesday:

All Traffic Sources

Top Content

(the rest were irrelevant)

Region

And for those interested in California…

Browser and Speed

So…

  • San Fransisco, Los Angeles, New York City, and Cambridge are good places to be. 40% of the visits were from outside the United States; Toronto, Bangalore, London, and Sydney were the popular foreign locations.
  • news.ycombinator.org?
  • It’s not clear where all the direct traffic is coming from. I thought direct traffic was made up of people typing in the URL, which wouldn’t be reasonable because it was so long.
  • 24K uniques/day? I’m not sure how to interpret the numbers. It would seem that only about 10% of the visitors to HackerNews clicked on the link to this blog. It’s actually even less because these numbers are over a two day period whereas the 24K number reflects HackerNews’s traffic in a single day. As noted above, a lot of factors could change the click through rate, but I’d still expect it to be a bit higher.
  • Why is it under Top Content the article has an average time of 7:33 (about what you’d expect for a post of this length), but under All Traffic Sources it says that the average time on the site was 55 seconds? I wish Google let you view the distribution so you could see, for example, how many people were on the site for less than a minute, how many 1-2 minutes, etc.
  • People still use Dialup and Internet Explorer?

If I missed any other stats that you’d like to see, let me know.

Six Lessons Learned from Domain Pigeon’s First Six Weeks

On January 29th I launched Domain Pigeon, a web site to help people find available domain names. The site lets visitors browse thousands of available .com domain names which they can register for their websites. It’s been more than six weeks since the initial launch and I wanted to do a recap of how things have been going as well as write about some of the lessons I’ve learned from the experience.

By the Numbers

Domain Pigeon has received more than 18,000 visitors and more than 62,000 page views since its launch. The average user spends about three and a half minutes on the site and looks at 3.4 pages. The site has a 45% bounce rate, meaning about 55% of the visitors look at more than one page.

The traffic looks like a positively skewed distribution:

The post-launch spike was a result of traffic from Hacker News, CNET, and ReadWriteWeb (RWW).

Currently the site receives about 1,700 visitors a week. The majority of the visitors (55%) still come from the CNET and the RWW articles; about a third come from direct traffic and the rest are from search engines.

The most important statistic, however, is the number of domains that have been registered. So far more than 339 domains have been registered. GoDaddy takes the majority of those with 177 out of the 339:

Lessons Learned

Have a Flexible Business Model

Domain Pigeon originally made money in two ways:

The first was paid accounts, which let account holders access four times as many domains as guests to the site. At the time of the launch this meant that account holders could browse about 12,000 domain names and guests could see about 3,000.

I had my doubts about whether people would sign up. I hoped that people would see that the registration fee was a small price to pay if Domain Pigeon helped them find a good, cheap domain name for their website.

During the planning phase I set the price of an account at $9.99, but I decided anyone who would pay $9.99 would probably also pay $14.99 and the loss in customers would be made up for by the extra money. My friend Rob suggested that the price end with a 5 instead of a 9 because it looked less like a scam. $14.95 it was.

The second revenue source was affiliate revenue. I signed up for an account with Commission Junction, which GoDaddy and a few other registrars use to manage their affiliate programs. The idea is that if someone goes to Domain Pigeon, clicks “Register” on one of the domain names and then follows through with a payment, I get a percentage of the sale. I had no idea how much money this would bring in, but didn’t expect much.

No one signed up for an account on the launch day, which worried me a bit. Fortunately, the next day I was hacking away when all the sudden I saw an email pop into my Gmail account: “Notification of payment received” Ha! I emailed my wife, who was at work, a quick message touting this momentous event. A few minutes later another email came in. Someone else had signed up. A minute later another one. “3!” I exclaimed in another email to her… What the hell was going on?

It turned out that Josh Lowensohn, a CNET blogger, had (I figure) seen the HackerNews post and written about Domain Pigeon. Not only had someone on CNET written about Domain Pigeon (and this is where the crazy optimism started), but all of the sudden people were signing up for accounts.

The next few days were great. The site was receiving a lot of traffic and people were busily signing up for accounts.

A week after the CNET article came the ReadWriteWeb article. During the three days after the RWW article Domain Pigeon saw 6,400 visitors, 13 account sign ups at $14.95/account, and $275 in affiliate revenue. The site had become profitable and life was good.

At the time visitors could register a domain through one of six registrars. After the surge in traffic I took a close look at which were bringing in revenue and GoDaddy took the cake. Something like 95% of all the registrations were through them despite having six to choose from. I decided to simplify things for users and to focus exclusively on GoDaddy, hoping that minimizing the visitors’ options would make life easier for them and increase revenue for me. That was change #1.

There were some clouds on the horizon. I was adding several hundred domain names to the site every day and I hadn’t considered the long term implications of doing this. With 3,000 domains available to everybody and 12,000 available for account holders, there was some incentive for people to sign up for accounts. But what would happen when the site had 100,000 domains available to everybody and 400,000 available to account holders? Would people still sign up? I’m not sure.

At the time I had made about as much from accounts as I had from registrations. Here’s the million dollar question: If I made $300 when the site only had 3,000 domain names available to everybody, would I make $1,200 if it had 12,000 domain names available to everybody? Almost definitely the answer is no, but, how much different would it be? Would I have made $800? $600? And if all the domains were available to everybody visiting the site, would that lead to more word of mouth advertising and more traffic and registrations? Hard to say.

I decided that charging people for premium access to domain names would hinder growth and so on February 16th I made every domain available to the public and stopped charging for accounts. That was change #2.

Not everybody was happy with the move. Several of the people who had signed up for accounts emailed me, all essentially saying the same thing. “So… why did I pay for this again?” I explained that the site was undergoing  changes and gave them refunds.

As it turns out, transitioning from a free to a paid service is a lot easier than going in the opposite direction.

Even after I made this change, not everything worked out the way I thought it would.

By the time I made the domains free to everyone, the traffic had died down so not as many people were making big purchases. I think most of the traffic that stuck around after the initial surge was comprised of collectors who normally don’t spend more than the small registration fee. They’re also not the types who buy hosting or certificates or anything like that. Additionally, since most of them buy in bulk, they don’t use the affiliate links, meaning I get no commission for the sales. Apparently affiliate marketing doesn’t work that well when you don’t get a lot of traffic.

Realizing this, on March 2nd I added paid accounts again, this time targeting the collectors. With an account, members could sort by the number of Google search results, view domains scheduled for the upcoming release,  and track domains by adding them to their favorites. It’s not much, but its there for the people that could use it. That was change #3.

It’s hard to predict how your visitors are going to respond to your site. I had worked out how much I would make with X account registrations, but I didn’t factor in how much I’d make from affiliate revenue and how those two would interact together to change my approach. Accept that change will happen, analyze the numbers, and adapt accordingly.

Market Along the Way

Say you have something big planned for your site. We’ll call it Shiny New Feature.

Should you spend time marketing your site when you know you’re going to add Shiny New Feature in a few days? If your marketing works and it brings in new traffic, will the addition of Shiny New Feature have a significant impact how they view the site? In Domain Pigeon’s case, that becomes: Will more visitors register  domain names and sign up for accounts after I’ve added Google sorting/layout changes/usability improvements?

I chose to wait until I implemented Shiny New Feature. The problem was as soon as I was done with Shiny New Feature, I found something else to do, and so I waited again. When I checked that task off the todo list, there was something else. “As soon as this is done, “ I said to myself, “I’ll start getting more involved in domain name forums, I’ll finally make that YouTube promotional video, etc etc.” So far, I haven’t really done anything.

In one sense, improving your product is a kind of promotion. The better your product, the more people talk about it and the more traffic it gets. But, it’s a mistake not to market along the way. If you have a good product you will constantly be iterating as you make improvements. If you wait for a final product, you’ll never wind up doing any promotion. Plus, the more eyeballs you get on the site, the more information you’ll have while determining what to work on next.

Substance is More Important Than Design


I spent a lot of time prior to launch working on the site’s layout. I felt that a well designed, usable homepage would have a tremendous impact on visitors. Believing this, I spent a lot of time tweaking font sizes, adjusting colors, and repositioning elements until I finally had a site that I was proud of.

Strangely enough—and this is symptomatic of the problem—I was more concerned about what people would say about how the site LOOKED rather than what they would say about what it DID. And so, that last Thursday in January, I posted a link on HackerNews and asked for feedback.

Their comments were, as always, constructive and insightful. Though there were a few recommendations on some usability issues, most of the feedback was focused on ways I could make it better for finding available domain names.

After the launch I spent even more time tweaking the design. Should the font size for this paragraph be 1.2em or 1.3em? #222222 or #2A2A2A? How could I reword this section to make it more concise? Does making the padding on this element 15px look better than 10px? Should I say “unregistered domains”, “free domains”, or “available domains”? And so on.

I spent the better part of February—the honeymoon stage–making small adjustments like these. I did make a few substantive additions, but I was mostly focused on small design issues. I wanted the site to look perfect.

At some point near the end of the month, frustrated with the decreasing traffic, I took a few evenings off to clear my head and to take a critical look at the situation.

Then it hit me.

A good design can greatly enhance your site, but it will not make it into somethings its not. Like makeup on a woman, it should enhance what’s already there. Design can’t make a bad site good and makeup can’t make an ugly woman beautiful. That is unless you’re drunk on optimism or alcohol.

You shouldn’t get too obsessed with design. Focus on things that will improve your site’s ability to meet its objective. That may sometimes include design, yes, but, don’t neglect content.

During the weeks leading up to the launch and for a few weeks following the launch, I had spent about 80% of my time on design and about 20% on substance. It should have been the other way around.

Had I realized this sooner, I might have launched a few weeks earlier. What’s the worst that could have happened?

People say…

Bad idea, great design – You stop working and move on.
Bad idea, bad design –  You also stop working and move on, but at least you didn’t spend all that time on the design.
Great idea, bad design – Now that you know your product is one that people want, go ahead and make it look good.
Great idea, great design – If you’re lucky or talented enough to be in this category, go after it with everything you’ve got.

I shouldn’t have been so concerned about what the people on HackerNews would say about the design. Criticism of your design is not a bad thing IF you have a useful product.  Had I known this I could have launched a few weeks earlier and spent that extra time making the site a better destination for people looking for domain names.

There is one downside to this, but I don’t think its as important as people think. If the site looked terrible, its unlikely that it would have been covered by either CNET or RWW. Without their coverage, the site wouldn’t have received nearly as much traffic as it did. But, does that traffic really matter? Don’t get me wrong: traffic does matter, but does THAT traffic really matter? Well, to answer that question, you have to know your audience…

Narrow Down Your Audience

It’s critical to determine who you are building your site for.

Until recently, if asked, I would have said that Domain Pigeon was for “people looking for domain names.” While that may be true, its so broad that it may have actually been harmful. It’d be like Walmart saying that they’re targeting “people who want to buy things.”

There are different types of people looking for domain names. Some have never created a website before and are looking to start one for the first time. Others are professional collectors, individuals who buy and sell domain names for a living.

Who was a I targeting? I never narrowed it down and it wound up derailing my efforts.

If I’m targeting people who have never set up a website before, I should include some basic tutorials to help people get started. I shouldn’t use the word “registrar” because someone who has never had a website is unlikely to know what that is. If I take this route, I should also should make the site completely free because beginners are more likely to click through to GoDaddy and to make a big purchase. They’re unlikely to sign up for the advanced features that might appeal to collectors.

Alternatively, I could target the collectors. In that case, I don’t need to include any tutorials. I could make the search feature interpret regular expressions. Beginners won’t need it, but a technical audience might take advantage of it. If I go this route, I should consider charging for access to the domain names, as I won’t make as much off the registrations, but they’ll likely be a significant percentage who will pay for a quality available domain list.

Because I didn’t narrow down my audience I have this kind of multipurpose frankenstein. The traffic is likely made up of collectors who came across the site by way of one of the blog posts. They get free access to thousands of domain names and they can even sort by the number of Google search results. The beginners—the people who have no experience setting up a site—don’t know about the site because its only been discussed in the tech world and even if they did, they wouldn’t know what to do with it if they did find it.

When I said that the the post-launch traffic might not matter, I meant that it depends on your audience. If I made Domain Pigeon for professional collectors, it really wouldn’t matter how much CNET or RWW covered it. The majority of the traffic would look around the site for a few minutes, leave, and never return. What matters, in this case, is how useful the site is to the professionals. I want them to come back; the RWW crowd is nice too, but they’re not longer term customers.

Your audience determines your business model, your layout, and the features you implement. Take your time to narrow it down and ask yourself if your site is best fit for their purpose.

Sleep

It’s 11pm and you’ve got a decision to make. Stay up and program or hit the sack?

When you only get to program two or three hours a day, the extra hour or two adds up to a significant amount of additional time in the long run. What might have taken six weeks might now only take four. The months are flying by so you better not waste it. Who cares if you’re a little tired the next day? If you don’t put in the extra hours, you’ll never get there, where ever there is. So you program.

Five or six hours later you wake up and you feel like hell. To your wife’s eternal dismay, you hit snooze three or four times and when you finally do get up, shower, and start driving to work you realize you forgot your phone or wallet or whatever.

The day drags on. You feel like a zombie and all you can think about is sleep. Finally, when the work day is over, you head home, prepared for another long night.

When you finally get back to programming again your mind is slow. You make typos. You get distracted easily. You look for excuses not to write that next line of code.

What you don’t realize is that that extra two hours you got in the night before has just messed up the entire day. Not only does the other 70% of your day suck because you’re so tired, but when you do get around to programming again you’re so slow that you barely make any progress. Ironically, the combination of mental slowness and typos might actually result in more work in the long run.

I’m a hypocrite for saying it, and I probably won’t even follow this advice, but sleep. You might not get in as much time at the keyboard and your project might take longer, but you’ll be more productive and you’ll feel better too.

Determine Beforehand how you Measure Success

Domain Pigeon was born out of a desire to learn something new and to prepare me for a startup one day. It’s a nights and weekends project that was never supposed to turn into anything serious.

After the CNET and the RWW reviews a crazy thing happened. It started making money. And the more money it made, the more I cared about how much money it made. It wasn’t a quick change. At first, I’d say to myself “This looks bad. It’s going to drive away customers. I need to fix it sometime.” But as the days went on all I saw were the blemishes and that’s all I wanted to work on. It didn’t matter that there were a ton of things that I had yet to learn. If the site didn’t look good it wouldn’t make as much money and if the site wasn’t making money, it wouldn’t be successful.

Ask yourself why you’re doing what you’re doing.

Money? Independence? Recognition? A challenge? Purpose? Enjoyment? Learning?

There’s nothing wrong with any of these—figure out which are important to you and measure your success based on how much you’ve achieved those things.

It’s easy to lose focus. The best way to stay the course is to determine what you value and stick to pursuits that help you achieve those things.

Use Your Product

Stupidly, I never really used Domain Pigeon to find domain names. While I was so busy building a site to help people find available domain names, I wasn’t in need of one myself.

About a week ago, just to see, I started exploring Domain Pigeon to see if I could find a good domain name for a future web app or company. Something short, pronounceable, and memorable. After about of an hour of searching, page after page, I figured out a clue as to why Domain Pigeon’s traffic wasn’t growing: most of the domain names are terrible.

From the homepage as I write this: bicuspi.com, icabbages.com, harmonisati.com, eearldom.com, westsu.com, laddertourna.com, uppercru.com, hyperbolicfu.com.

Unfortunately, its not an easy fix. Algorithmically determining which domain names are good and which are bad is a tough problem. Most of the obviously pronounceable ones, like words in a dictionary, are registered. Even if it could be done, tastes vary. Examining some of the domain names that have already been registered show names like “bezzl.com”, “circleregister.com” and “masterofthe.com”. Would you rather see 10K domain names that have a high probability of being good or the same 10K names mixed in with 90K names which are likely not good, but hey, you never know?

Looking Forward

There’s a lot of things that can be improved with Domain Pigeon. There’s too many bad domain names and too few good ones, there’s no way to generate domain names around certain words, and there’s too little traffic for the colors to be relevant. With some more work and a bit of marketing, who knows what will happen. :)

Entrepreneurship is Getting Easier

Really interesting article in the Economist:

The triumph of entrepreneurship is driven by profound technological change. A trio of inventions—the personal computer, the mobile phone and the internet—is democratising entrepreneurship at a cracking pace. Today even cash-strapped innovators can reach markets that were once the prerogative of giant organisations.

Another reason for entrepreneurship becoming mainstream is that the social contract between big companies and their employees has been broken. Under managed capitalism, big companies offered long-term security in return for unflinching loyalty. But from the 1980s onwards, first in America and then in other advanced economies, big companies began slimming their workforces. This made a huge difference to people’s experience at the workplace. In the 1960s workers had had an average of four different employers by the time they reached 65. Today they have had eight by the time they are 30. People’s attitudes to security and risk also changed. If a job in a big organisation can so easily disappear, it seems less attractive. Better to create your own.