Yesterday was an administration day at infovark. I spent the whole day not writing code.

When you start your own software company, you suddenly discover the amount of time you can spend on things other than making software. Before Gordon and I made the leap, we created a high-level schedule to take us to our first release. If you count effort expended against the plan, we’re tracking well. If you measure by milestone dates, we’re far behind. And it’s all due to things other than writing code.

We drastically underestimated the amount of time taken up by finance, legal, marketing, administration and human resources. It’s an understandable mistake. At our previous jobs, we had people to look after these things for us. Now we’re in charge of them as well as heading up our R&D efforts.

For other software entrepreneurs, here’s a short list of things that aren’t code:

• Incorporating your company
• Setting up a bank account
• Conducting shareholder meetings
• Researching your competition
• Setting up your website
• Figuring out health insurance
• Building your brand: logos, colors, style, etc.
• Purchasing hardware and software
• Setting up your phone system
• Setting up email
• Buying business insurance
• Determining product pricing
• Explaining to all your friends why you’re doing something crazy like starting your own business

All of these things are important, of course. You wouldn’t have a company without them. Yet the developer in you desperately wants to be left alone to “get some work done.” It requires an attitude adjustment. As an ISV, it’s all part of your work now.

Before we officially launched infovark, Gordon and I read Eric Sink on the Business of Software. It’s required reading for all developers working in an ISV or thinking about starting one. You can also check out his blog. It’s a great place to start learning about all those important things that aren’t code.

And now back to the fun stuff.

Continue Reading

Dion Hinchiffe authored an essential post on the State of Enterprise 2.0. It’s less a survey of Enterprise 2.0 technologies or vendors than a survey of the state of adoption among mainline businesses. Some key insights I gleaned from the article include:

• Enterprise 2.0 is more about a change in mindset than a particular set of technologies. Enterprise 2.0 products are social tools, and cultural change accompanies their use.
• Organizations will find these tools proliferating throughout their enterprises, due to low barriers to adoption. Remember that much of Enterprise 2.0 was piloted on the Internet. The designers of those tools knew that the only reason their applications achieved wide adoption was because they were easy to use and solved customer problems. (Even if the problem is something trivial, like what my friends are doing right now.)
• Enterprise 2.0 technologies tend not to displace existing IT investments, but to add value to them. This is largely due to the grassroots adoption of Enterprise 2.0 tools, but also with making participation in online communities optional. Join if you want, contribute if you want, but nobody can force you to do so. (In fact, as the FASTforward blog points out, it’s rare to find firms that aren’t actively standing in the way of Enterprise 2.0 adoption.)

Head over to the ZDNet blogs and have a look.

Continue Reading

Radar Networks has announced their “first mainstream semantic web application”, called Twine. It looks to be an intriguing blend of some of the elements of wikipedia, facebook, and a sprinkle of digg and del.icio.us thrown in for good measure. You can see some screenshots, and get a better overview over on readwriteweb. (Twine is currently pre-beta)

It’s a hosted service, that allows for people to create semantic structured web content around themes and topics, called twines. (In some ways, it reminds me of squidoo, too.)

I’m not going to stick my head in the flamebox that is “the semantic web”(because I know Dean is working on that post), but I will say that I really like the way that twine is shaping up to manage relationships with entities other than people. If the semantic web can deliver this in a meaningful way, I think we’ll all be a bit happier.

(But the “Web 3.0″ moniker makes me cringe. I can’t help it. And I’m all about the buzzwords, too…)

Continue Reading

Gordon and I jokingly call infovark an “N-complete” enterprise. We do this for two reasons.

1. We’re nerds. We think about math stuff like np-complete problems for fun.
2. We use most of the open source N* tools available today in our software development process. We’re nerds.

This means NUnit, NCover, NDoc, and NAnt. For the non-nerdy, these are frameworks that help developers manage unit tests, determine code coverage for said unit tests, document our code, and automate our build process. We bring all the tools together using SharpDevelop. (Though I do miss Microsoft’s Visual Studio a bit.)

Why use all of these tools? As a small software development shop, the most important thing for us is to manage our time properly. As Joel on Software pointed out, multitasking developers face context switching costs. Revisiting our code after the fact to write automated test cases or add documentation is a huge time-wasting exercise. It’s much better to do these things while we’re down in the weeds, working with the bits and bytes. Having the N* suite running within our IDE makes us much more productive.

We think being our own testers and tech writers gives us better quality code as well. After all, the person writing the code is best equipped to describe what it does and the most likely ways it could break. It makes sense to do these tasks ourselves as long as we can minimize the switching costs.

Our thanks to all the open source developers that worked on these tools. We wouldn’t get far without you. And thanks as well to the FxCop team, whose code security tool is a real godsend. (But couldn’t you guys have named it N-something?)

Continue Reading

Well, it happened a lot sooner than we thought, but infovark has made the front page of Joel On Software:

Dean and I were at Crystal City yesterday to hear Joel talk about the new features in FogBugz 6.0, which looks like a great product.

My favourite feature was the evidence based scheduling, which uses the history of individual developers estimating accuracy to calculate a series of projected shipdates and probabilities, shown in a fancy AJAX chart. Feature/Schedule tradeoff decisions have never looked easier than they did on the big screen yesterday.

Kudos to the Fog Creek team for bringing a whole lot more reality into the software development process, and for building a system so easy that developers might actually use it.

Continue Reading