Kent Beck - "Before [market fit] you need the absolute lowest possible latency from idea through implementation to validation. ... Sometimes this means no autom

Kent Beck - "Before [market fit] you need the absolute lowest possible latency from idea through implementation to validation. ... Sometimes this means no automated tests." I think this is a great point and perhaps explains the low support for TDD on HN. In my experience the properties of a TDD project are: consistent flow rate of business value, low regression rate, low QA cost (assuming integration tests). It's the consistent flow rate which is perhaps most interesting. At the start of a non TDD project it's easy to achieve a high velocity without tests. As the length of time from the start of the project increases the velocity will inevitably decline due to the increased complexity of the project and the greater level of QA required. Over a longer period the project without tests is going to tend towards a slower velocity until eventuality it hits a wall and it's big rewrite time. For a TDD project the velocity curve is going to be different. You're going to see a slower start but once a velocity has been established it tends to stay fairly flat or increase. The XP guys talk about hyper-productive teams, there's no guarantee but you're far more likely to hit this point with a TDD project. Increasing the number of features a product has and the number of team members will only exaggerate these two velocity curves. There's another factor here as well, the knowledge of TDD within a team. The greater the TDD experience the smaller the difference will be in the starting velocity. With enough experience the velocity hit becomes small enough that the two curves can cross over very quickly. The choice of strategy then becomes: are the velocity curves likely to cross over before or after you achieve market fit? Not an easy question to answer and one which is easy to ignore when everyone is excited about just getting something built. edit: not sure what happened to the parent post, here's a link to Kent Beck's comments about testing before market fit http://www.quora.com/What-kind-of-automated-testing-should-a-startup-have-from-the-beginning-through-the-first-six-months-of-live-operation http://www.quora.com/What-kind-of-automated-testing-should-a...