I was listening to this Swedish podcast where the episode (episode 8) was titled "stop estimating" and since I hate estimates I started listening with high expectations. Naturally the title was chosen to be provocative but I was surprised that the message was actually to stop estimating.
There are three important messages here I think. First it is that estimates are useless. or at least the type of estimate most engineers do. I completely agree here. You are always asked to make estimates when you know the least about a problem so by definition the estimate is going to be terrible. So the value of the estimate is very low. If any value at all.
However there is always important to understand what is expected of each requirement, user story or whatever you are using. I sense that the people in the podcast are used to teams that are good at breaking things down into easily understandable chunks while my experience is that that is rarely the case. In this situation estimating (the verb and not the noun) can be a useful tool. The process of trying to come up with an estimate can be a great way to force a team to understand what each user story is about. The actual estimate however can be thrown away if you ask me...
The last important message is that if you are estimating, maybe for the reason described in the previous paragraph you need to focus on relative estimation rather than absolute. And as said in the podcast; don't estimate your complete backlog. You only need a handful number of tings estimated, the next things you'll work on. Leave everything else for the future.
Anyway if you know Swedish I think you should listen to more episodes of Väg 74.