You're right about those user estimates and various knobs having a potentially huge effect on results. When I introduce a Monte Carlo simulation mode, I hope to

You're right about those user estimates and various knobs having a potentially huge effect on results. When I introduce a Monte Carlo simulation mode, I hope to be able to offer better insights into how distributions around some of these variables will impact likelihood of success. I'll also continue to look for opportunities to create smart defaults and/or links to resources that can be used to help calibrate estimates based on historical data, locality, etc. Thanks for the comment!