Gyaan Sutra

How long does it take to completely revamp a web application?

Mbakarma needed a few major changes and I’ve been waiting for way too long to do them. So last sunday, I decided to start work on it again. I wrote down the features, walked through the details, and finally decided on a list of features and improvements which I think are essential.

Now I didn’t want to spend too much time on it, because you know you can spend forever with it and it would still not be complete, so I chalked down a small timeline in which I could add everything I needed.

To make sure I’m doing things on time, there are a couple of really cool applications I’m using this time.

One is Bean Stalk App, which is an online SVN repository. And with a few clicks it integrates with Twitter and the other project management application that I’m using. I’m still using the free account on BeanStalk, and if it all goes well I’ll shift to the paid account soon. If you are a startup and thinking about starting work on your application, I would suggest you get an account right now on beanstalk and start using it. You can try running with a local svn repo, but you know hard disk failures are not that uncommon. Also, BeanStalk provides you with fast backups, so that in case they have a hard disk failure, your data doesn’t get lost.

The second application is a really a masterpiece. It is called FogBugz. It uses evidence based scheduling to calculate how long it would take to complete the project. Behind the big name, what it really does is this - It asks developers to estimate how much time would each feature/bug would take to complete. Once you allocate a time, and start racing against the time, Fogbugz tracks how much longer you took compared to the estimate you had given. After you’ve completed a number of feature/bugs, it calculates the probability in which you would complete the fix. So if you always take double the time, and you estimate 2 hours, Fogbugz will count that as an estimate of four hours. Once you add all the developers working on the project, and their probabilities, you have a probabilistic model of how long it would take you to complete the project. There are a number of other really awesome features there, so do check it out when you have the time. They have a 45 day free account, which is long enough for you to find out if you like it.

Still, those are just accessories. You can’t rely on them completely and so you are still very important in the whole process. My overall estimate, which I’d rather think of as saying that it is my word, that I would have completed everything by the 15th of April. I know we are all suckers when it comes to estimation, so let’s just see if I’m able to stand by my word. After all, what distinguishes us as humans, is our ability to keep our word.

If you want to keep track of how well I’m going with my estimates, (so that you can blast me later) you can follow me on Twitter.

Also, once I’m done, I’ll be back with good long posts about what I like and what I do not like about these two applications, and what hot and happening in the brand new Mbakarma.

Till then, Cheers.