@Rawalanche,
I totally digg ya but let me try to rephrase my point in a different way so maybe you'll understand what I am talking about.
Say for example, you are rendering a frosted glass region which takes a lot of render power to get to acceptable passes. With the option of having the bucket mode I can set the number of samples to (about)50 and let it start rendering through the region. After 2 buckets I'll probably see if my reflection color, glossiness and all of that basic stuff is good and set. At that point I can cancel it and either fix it or switch to the full progressive mode.
If I want to see a larger portion of the region I'll just let it render to about 50% of that render region. By that time I'll have a clearer idea if my material is gonna work or not. If not, I can render it in its entirety and switch to the progressive renderer the next time I want to see a preview.
On the other hand, if I did a progressive render region I'd need to wait a lot more just to have an idea of how my glossiness is working (for example, right?). That, or set two different regions.
Yeah, rendering the whole render region with adaptivity + denoising would be faster but sometimes you don't need that. And if you do realize you need it, you can let the buckets finish and then for the preview you switch to the progressive renderer.
You can also take into account how buckets can be helpful at 5k and super precise displacement, meaning I can quickly see the end result of one bucket and the ability of having the option to let it continue rendering through the whole region without canceling the entire render just to re-set the render region and the render engine calculate all of that displacement again.
I'm just saying, some of us are apparently used to working like that. For some perception comes into play as well (seeing the final image without the progressive passes)... I mean, to each his own really. Personally, I really like both modes :)
I guess it is harder to explain that I thought, ha!