Time estimation of software development should be a product of observation of historical evidence, and many factors that come with it, like:
- What was the language used?
- How many developers worked on it
- How many years each developer has in experience in programming?
- IQ of each developer
- How many kids they have
- The weather
- ...etc
Analyzed by data scientist.

TL;DR
Not something you get by asking developers and interrupting their work, because many are people with superior complexity who often overestimate their capability of solving given problems.
Don't trust them to estimate!