This is a great qestion!
Dependent claims are such a prominent thing and they don't give you anything in regards to claim scope. (Apart from interpretation maybe, but a well written description could get you there too.)
But let's look a little closer. (I'll be citing the epo manual because I know it better, but it's pretty close to the US thing)
Drafting the patent:
Traditionally, the description will be drafted as to enable the claims. It's very hard to write a description without dependent claims. Sure, you could remove them afterwards, but I'll get to that.
The dependent claims are something you have in front of you - in the form of a claim - so you can make sure it's clear and precise and consistent. During prosecution, it is always harder to enter something from the description, because the description is in, in lack of a better word, description format and you might run into an (intermediate) generalisation.
So, writing good dependent claims which may become independent later is the basis for a strong patent. in this stage, it's only about having thought them through, you could now throw them away theoretically. Which leads to:
The claims will be searched by the examiner (as long as they relate to one invention). If you have no dependent claims, it's at the examiners discretion to search all those tiny bits and pieces from the description (spoiler: they won't).
Trying to include limitations from the description can then lead you to unsearched matter, which at the EPO means you have to file a divisional or not include it, because they won't examine (and therefore grant) anything not searched.
Additionally, the above as to generalisations and in general missing support for the new claims applies, but that could be remedied by a well drafted description as I said.
So let's say you got through prosecution, why not remove all dependent claims now and not pay claim fees?
Or rather, invalidation. Funny enough, what I said for the phases above still holds, let's say your independent claim is invalidated. You now want to add a limitation from the description. But oh, it wasn't searched, is a generalisation or not clear? Too bad.
But there's another point. Litigation goes: "opponent files an invalidity suit/opposition/... - responses - oral proceedings". Somewhere along those, you will have to add auxiliary requests with new independent claims which you can then go through with your opponent in the oral proceedings. Your opponent needs to be prepared for all your dependent claims and the axuliary requests in the oral proceedings. Not for you entering stuff from the description. So the proceedings will be a lot longer if you keep adding stuff from the description.
And it might even brake your neck in the end, because Art 84. (EPC), also called "clarity (and brevity)" is not a reason for opposition (at the EPO). If you got your dependent claims through the examination, they are presumed clear! If however you enter limitations from the description, your opponent may argue missing clarity.
Now, you'll have to believe me on this one, examiners almost don't care for clarity. Many many many granted claims would be found unclear after thorough dissection by your opposing attorney. But they may not argue that for the existing claims.
And, did I say that was it? Well there's more.
Interpretation in invalidity and litigation:
Let's say the examiner granted your claims. Some of the dependent claims (and the independent one too then) would be non novel over the prior art which the examiner examined in a special, but possible, interpretation. The EPO grants the examiner a presumption of having done his work well, so in most cases, your claims will be interpreted to not contain that interpretation. So they aren't invalid and you can actually use them. Technically, the independent claim would be enough, but maybe that one fell against a new prior art document and etc. In the end, the dependent claims allow many arguments as to what is part of your claims and what not, most of them good for you as the arguments against you could almost all be made from the description anyways.
The only case where you might get a problem is if you include a dependent claim with an embodiment which could have been interpreted to be outside of the independent claim and is not novel. Then your opponent can argue that anything in the dependent claims is definitly part of the independent claims. That argument is harder if the description doesn't clearly state it is part of the independent claim. But again, in that case you would also not be able to include that limitation into your independent claim if you had it only in the description because that would be unclear. So this one is an error in the drafting of the patent.