At the beginning of every iteration our whiteboard is covered with "not started" stories estimated in IDEAL HOURS. On a daily basis we update these estimates to keep track of each task's progress. I think we are doing it wrong. The original estimates are in ideal hours and our daily "re-assessment" in ELAPSED TIME. This seems inconsistent, right?! How would you measure your progress towards the end of the sprint?
Reestimating tasks during sprint is non-sense. It will just spoil burndown chart and it will not change commitment you already did.
Anyway in my opinion whole point of "estimating" tasks is waste. What you will get by estimating these tasks?
- Will the estimate will make your product better?
- Will the estimate increase the quality of the product?
- Will you make your product faster if you spend time on doing estimates?
- Will your product be better if you do more estimation?
I don't think so => task estimation doesn't bring any value.
Now what will happen if you don't do estimates?
- Does it mean that you will not be able to track the work progress?
- Does it mean that you will not estimate?
- Does it mean that you will not be able to do commitment?
No because tasks are created from user stories which are already estimated. Commitment is done on user stories so it doesn't matter what is estimate of tasks. Only important thing to track during sprint is amount of completed user stories which can in case of burndown chart still be represented in amount of delivered "story points" or any other unit used to estimate user stories. Tasks should be small and it doesn't matter if small mean day or 3 hours. Developers should be skilled enough to create such tasks without need for estimation. It means that task estimation doesn't remove any value.
If we go to lean principles any process which doesn't bring any added value and can be removed without loosing any value is a waste and should be eliminated.
But perhaps you get some value from task estimation - every team can implement Scrum little bit differently. If you doubt you can use great feature of agile. You can simply try it in one sprint and if it doesn't prove itself useful you can return to task estimation.