technology time estimates

Wei Dai (weidai@eskimo.com)
Mon, 12 Jul 1999 12:04:11 -0700

On Mon, Jul 12, 1999 at 10:53:50AM -0700, hal@finney.org wrote:
> It's interesting to consider how we should update our time estimates
> when earlier estimates are wrong. Should we take into consideration the
> earlier mistake and use it to increase our current estimates? Or should
> we just ignore the past and do our best with the information we have
> at hand?

We should figure out why we apparently consistently underestimate the time needed to develop techonologies. Is there a real systematic effect, or do we just pay more attention to the underestimates than to the overestimates? If it is real, what is causing it and how can we avoid it? Are there any studies on this subject? Maybe someone can do an analysis on a bigger (more representative) set of IF data?