Based on my 34 years of getting paid to make games, at about a dozen different companies, 50% of the time when someone gives an estimate it is just plain wrong. The longer the task, the more wrong it tends to be and often massively wrong. The other 50% of the time they may be close or not too far off, but still can be off quite a bit. When wrong it is almost always underestimating. I have worked with guys who I learned to just x3 however long they said anything will take, which is my rule these days for even my own tasks.
Game development is full of unknowns. How do you estimate time of an unknown? By finding something close and trying to remember how long that took. Good luck, as most tasks were not done in a linear fashion but mixed up with other tasks.
Smells is like noise, but different, but wait the new smells is copying parts of noise, so does it even matter how long noise took you if you could even figure that out as we are not starting from square one. Months ago I told people, hey I could make it like noise and have it working in a day or so. That only gets your foot in the door, so really several days, then others wanting this change or that and hours of talking about it and more changes and has probably been a week of work so far of just my time, with Allan also roped in for hours of xml changes and testing.
A method you could do is say nothing until a release is a few weeks from being really done, then announce you will release it 3 months from now, which should mostly guarantee success, if there were such a thing. I think some companies do that, but it is really hard to be that patient with all the pressure to build hype and get products out the door.
And now my 5 minute drive by estimate is way off, but my wife is sleeping, so don't tell.