How do you determine how much time is needed for a task?

Question

Programming

No matter how much I worked with sites, estimating the time needed to solve a particular task was the biggest problem.
It is worth mentioning right away that we are not talking about understandable tasks, or tasks that have already been done 100,500 times before - everything is clear here.
But what if you don't know how long a task will take? For example, they let me work on a new big project, written on a crutch, they ask you to fix something there, and they immediately ask you to announce the time.
How will you do it? What will you answer?
Of course, you can break the deadlines - voice weekly and monthly deadlines even for small (seemingly) edits. And to a large extent it will be reasonable. But personally, my conscience does not allow me to do so. Could this be the problem?
In general, I would like to hear opinions on this topic again.

Answer 1

1. We estimate the complexity of the task based on the assumption that everything is clear to us, there will be no unexpected problems or additional wishes.
2. Multiply by pi.

Answer 2

Obviously, if it is not possible to evaluate immediately, then the task is non-trivial and requires time to study and evaluate. Therefore, I say that I need time to understand what and how to do and how long it will take. Of course, we add some time to the final assessment, just in case.

Answer 3

This is already from the field of software engineering, different agile and non-agile methods. Take for example Scrum or XP with their story cards, planning poker. Example:
- There is a big task. You break the cards into stories, for example, the user enters his data and clicks on the registration button, which is also further divided into tasks, which ones can be seen here? The user enters his data? What data? - First name last name address phone. What happens when you press the button? The data is taken or written in db, from this point you can also do something like "database connection". And in the end, when these pieces of sausage are the smallest cut into small pieces. There is a distribution of time for each of them (by the way, they can immediately determine the time for the entire user story). Defined as Story Points, where 1 = full slave. day (may vary). And so the team sits and allocates time for these tasks, for example, I say that to connect a db (now it’s very abstract !!) you need 0.5CP, Vasya says 1CP, and Petya says 0. 1CP in the end it turns out that 3 people from the team do not agree. Then each of them begins to tell everyone his point of view, why he thinks so, why exactly 0.5 or 1 or 0.1 everyone listens - and vote again based on what they heard, until the time matches or at least approximately.
After that, each user story has its own total time weight. They all add up. And now, please, I passed the time analysis (in Scrum, it is really done for 2-4 weeks (1 sprint)). But in practice, many make mistakes in the calculations, and therefore do it every month, thereby increasing the accuracy of these predictions.
And it is very difficult to calculate everything at once and it will not be accurate. If the customer is adequate, and in theory it should be adequate, since we are talking about a large project, then explain to him using more understandable examples (on the fingers) that things are not done this way, and if they are done, then you "f**k* t".

Answer 4

It's simple:
1. the first assessment - for the study of the issue, time is limited, usually a few hours (2, 4 or 8)
2. the second assessment is possible only after the first part - the study.
In conditions of high uncertainty, nothing better was invented, the first step is to reduce uncertainty. Obviously, research time is also paid.

Answer 5

It comes with experience. The more such tasks are done, the more examples for comparison.