Joel, in Person, on his Software
Software
Joel Spolsky brought his FogBugz tour to the DC area on Tuesday to demo the latest version of Fogbugz and talk about software. Before it started the omens where good; the guy in front of me walking in had a CFUnited backpack on and the audio system was playing the Dave Matthews Band.
Fogbugz 6 brings two new features to the product; a wiki and Evidence-Based Scheduling in an effort to change Fogbugz from a bug tracking tool to a "complete project management system". I don't, however, want to give a review of the product but rather give my impressions on what Joel Spolsky said about software and Fogbugz. Some notes;
- The idea behind Evidence-Based Scheduling (EBS) is an intriguing one; take existing data of how accurate a developers task estimates are versus actual task time, weigh it over time and then use it to give a probability of project ship dates from 0% to 100% chance. Having seen, and worked with, gant charts and other such things with varied success this certainly seems an approach that is worth a shot. Most of the discussion centered around this new feature and it provided an insight into Joel's philosophy on developers.
- Joel does not believe in project plans where task 71 is dependent on and can not start until task 62 is complete. Software is not like constructing a building and good developers will go ahead and create a fake query or put in place some scaffolding if a dependent task is not fully complete.
- From Joel's experience it takes a developer about 6 months to get accurate in estimating.
- Always estimate in the powers of 2; 15 minutes, 30 minutes, 1 hour, 2 hours, 4 hours, 8 hours, 16 hours. Anything higher than this and way too much is going on. I've heard this elsewhere and been using it as a guide for a while and it works well although I have never estimated a task for 2 days.
- Some in the crowd seemed to want to use the tool to find those responsible for failure or get rid of bad developers. It was clear that Joel thinks in a different way. He is not interested in blame but rather helping the struggling developer become a great developer. All through Fogbugz there is a sense of trust of the developer.
Joel is a very good speaker and has many insights into developing good software. Even if you are not interested in Fogbugz I would still suggest going to hear him speak or read his blog.
Fogbugz 6 brings two new features to the product; a wiki and Evidence-Based Scheduling in an effort to change Fogbugz from a bug tracking tool to a "complete project management system". I don't, however, want to give a review of the product but rather give my impressions on what Joel Spolsky said about software and Fogbugz. Some notes;
- The idea behind Evidence-Based Scheduling (EBS) is an intriguing one; take existing data of how accurate a developers task estimates are versus actual task time, weigh it over time and then use it to give a probability of project ship dates from 0% to 100% chance. Having seen, and worked with, gant charts and other such things with varied success this certainly seems an approach that is worth a shot. Most of the discussion centered around this new feature and it provided an insight into Joel's philosophy on developers.
- Joel does not believe in project plans where task 71 is dependent on and can not start until task 62 is complete. Software is not like constructing a building and good developers will go ahead and create a fake query or put in place some scaffolding if a dependent task is not fully complete.
- From Joel's experience it takes a developer about 6 months to get accurate in estimating.
- Always estimate in the powers of 2; 15 minutes, 30 minutes, 1 hour, 2 hours, 4 hours, 8 hours, 16 hours. Anything higher than this and way too much is going on. I've heard this elsewhere and been using it as a guide for a while and it works well although I have never estimated a task for 2 days.
- Some in the crowd seemed to want to use the tool to find those responsible for failure or get rid of bad developers. It was clear that Joel thinks in a different way. He is not interested in blame but rather helping the struggling developer become a great developer. All through Fogbugz there is a sense of trust of the developer.
Joel is a very good speaker and has many insights into developing good software. Even if you are not interested in Fogbugz I would still suggest going to hear him speak or read his blog.
Loading....