1. Kevin

    Your comment on variance and short/long term estimation is interesting to me. From a developer’s stand point, it seems easier to provide an accurate estimate for the short term, while the long term is difficult. But this point of view is likely highly bias based on my experience.

    • Shane Kercheval

      Kevin, excellent question/observation.

      Regarding estimation and short-term vs. long term. I definitely understand your point and, to be honest, i’ve gone back and forth on this. Intuition would tell us that it is easy to measure something short than something long. However, the key point relates to the variance of the estimation (i.e. how accurate it is compared to what we actually did), and how that might affect our efforts (overestimate the workload and you will pull stories in, so why did you need to estimate in the first place?; underestimate workload and it is possible there will be a big push at the end). For example, if we were just two days behind our estimation for a 2 week sprint (i.e. it took 12 days instead of our estimated 10, assuming there was no big push at the end to quickly wrap things up), then we underestimated by 20% ((12-10) / 10). Now, if we have a 6 month project, and we were 20% off (underestimated), then we would be over a month late (1.2 months). Obviously this isn’t exactly how variance is calculated mathematically, but it is very close to the same concept (percentages vs. absolutes) so (IMO, and from what I have seen), it turns out that it is much easier to stay within the long-term estimations than the short-term estimates in terms of percentages, and thus the variance tends to be smaller (e.g. our customers would not be very happy if we were more than a month behind our final deliverable, but is a big deal if we are 2 days behind a sprint? Probably not, since all the variances will average out). This absolutely does not take into account that there are still the common push towards the end of the project, and teams might cut/add stuff to meet the long term deadline, so there are definite arguments against my thoughts. So, basically, yes, it might be easier to estimate in the short run, but easier doesn’t necessarily mean more accurate, IMO.

      Thank you!

  2. Shane:

    Good points… the need to balance dynamic change associated with the customer and the environment is a driving factor…

    The type, quality and density of information used to manage this dynamic change is different in each of the reviewed process types.

    The combination of flexible production systems and change control systems based on high-density information artifacts tends to produce a more valuable system production process.

    Have fun,


Leave a Reply

Your email address will not be published. Required fields are marked *

2 × = fourteen