Historical errata for Software Estimation Without Guessing

Model-Based Estimation paragraph:

Direct comparison is the root of estimation, but it we’re going to…

Empire Enterprises section

Some of it handles the all of the corporation’s accounting processes.

Reference [1] shows ‘url>’ at the end.

“If we were talking about something repeatable,
then a 90% confidence level would indicate that 90% of the time the actual
would come in within the estimation.”
It appears to me, that a confidence level is not useful without a range.

Example for a useful statement: “I’m 90 % sure, that the work will take longer than 2 days but no more than 4 days Is”
Example for a useless statement: “I’m 90 % sure, that the work will take 3 days” (because the work will never take exactly 3 days).

“ estimating at the 50% confidence level is more appropriate. This means that half the time we overestimate and half the time we underestimate.”
I’m neither a native speaker and nor an expert in statistics, so I may be wrong. It appears to me, that this is the definition for the 50th percentile, not a 50 % confidence level.

»Since we don’t get to repeat the same software development project in the same circumstances to measure that, we’ll just treat that as a colloquial expression meaning “pretty sure.”«

I’m neither a native speaker nor an expert in statistics, so I might be wrong. It appears to me, that the confidence level tells something about repeated estimations, not only repeated development work. Isn’t the meaning of ”90% confidence level”: Of all the projects, for which an estimation was given with a 90% confidence level, the actual effort was inside the given range (!) 90% of the time.

Some of the burn chart images are in the wrong places.