A lot of things in the new API just seem like illogical choices to me... the first obvious one being the lack of seeking in the playlistitems end-point. I wouldn't want to do a token swap until the end, because that would be a huge strain on some people's servers. Sometimes it can take 2 full seconds for a cURL operation, if you have to do 10 in a row, that would be ridiculous.
Another odd choice would be the change of their "duration" tag from the standard seconds counter, to a very odd and non-standard text format of TS[d]H[d]M[d]S. It makes it so that ANYONE who wants to calculate timestamps, they must do a regular expression to actually get the required information.
Then of course, there is the hiding of keywords. You can no longer fetch keywords through the API.