I'm not as sure as Dan & Matt that this obviously makes more sense as part of urlBuilder than #-parameters. If assigning parameters from encoded strings, such as script parameters and result, to variables is part of the purpose of #-parameters, isn't this consistent with that purpose?
On the other hand, #-parameters is designed to be a coherent set of functions operating on one data format, whereas this would be a new data format. Dan spun conversion between JSON and Let Notation off as a separate module rather than making it part of #-parameters, and I think that was definitely the way to go for that case.
I haven't tested this, but I presume that URL variables also wouldn't be capable of the data type preservation feature of #-parameters. If scripts are accepting parameters either as Let Notation or as URL variables, any scripts sensitive to the data types of the incoming data will have to coerce data type to accomodate the URL variables, which renders the type detection in #-parameters redundant. So maybe losing that feature in format conversion is ultimately less of a concern anyway.
Would it be so much of a burden for scripts using Let Notation parameters to be called with "param=" in their URLs instead of variables? Sure, that will make the URL more verbose, but so what? The URL and parameter values will be less human-readable, but URLs are pretty difficult to read in the first place. How frequently are we sending so many parameters that are so long that we run against the limit of how long URLs can be? I ran a test a few weeks ago, which was limited to intra-file URLs running locally on a Mac, and I crashed FileMaker before there was a limit on how big the URL could be.