Ask any startup. The most difficult decision leading up to a public release is when and what? Some might argue that getting funding is the most difficult but a good startup avoids funding until later, if at all. Others might argue that the difficult part is getting the right mix of people and hitting milestones. That also is important, but not as important as the when and how.

Usually, a good launch product is the result of a perceived need. Or maybe a need not yet realized – it’s hard to say for sure. There’s some black magic involved in all that.

FriendFeed launched not long ago because there was an empty hole in Twitter – that was aggregation and conversation. FriendFeed figured out that, to be successful, it was going to target that emptiness in the highly popular Twitter experience.

Disqus and Intense Debate figured that, in order to be successful, they needed to target the missing piece in blog comments – that was reputation and reputation management across blogs. The two fight it out, post-launch, over which is going to differentiate it over the other.

In these cases, the timing of the launches was critical to the uptake. Twitter started experiencing significant problems and influential early adopters began getting itchy to be somewhere that scratched their itch.

Putting aside timing, the most important part of a launch is what. It’s feature-sets. It’s determining the balance between a fully developed roadmap of features and what is needed to “hook” early adopters and get them to stay.

Take Identi.ca, the new Twitter clone that is completely open source and is timely in that Twitter faithful are really, really close to burying the hatchet and simply abandoning it altogether. The timing could not be more perfect. Folks have been talking about distributing Twitter and relieving the strain of a centralized service at one time. Open sourcing the product does this, to a degree.

However, Identi.ca gets a big “FAIL” for its launch for a few very important reasons.

1. There is no coherent way to deal with “replies”. Folks used to Twitter realize that when there is a river of content, and that’s what Twitter is, there must be a way to manage conversations. There must be a way to keep up with followers who are talking to you. In my working with Identi.ca, there is no way to do that and, while that might be coming, it wasn’t there at launch. Very conceivably, I’ve been lost forever and I generally have tons of followers as an early adopter. FAIL.
2. XMPP doesn’t work. The one reliable way to reply that folks on Identi.ca were talking about last night was with XMPP, the protocol used for various IM clients including Google Talk. I could deal with replies that way if it worked but at some point, XMPP stopped working. I could receive, but I could not send. A one way conversation is a monologue. FAIL.
3. OpenID integration must be seamless. I was pleased to see OpenID supported when I signed up. Unfortunately, today, I could not login with my OpenID account. If I can’t get in, I can’t use it. FAIL.

Some would say I’m being too hard on this startup. Screw that. Perform or get off the stage. There are very obvious and defined features that must be included in a microcontent site at launch. I’m not saying an entire roadmap needs to be worked out. No, get a working beta up and get testers in there. However, without replies, without reliable “offline” access (i.e. IM, SMS or client integration) I’m not going to stick around. Finally, direct messages would be a nice feature.

While I have high hopes for Identi.ca, I will remain there only to squat on the name “technosailor”. Bye, guys.