The ISP's are not responsible for shaping the connections (except for FTTP, but if they don't NBNco has policers that squelch the connection, so it ends up slowing to a virtual halt with high packetloss. Net effect is non-working connection, so if the ISP was failing to shape the connection would not work).
FTTN specifically, the line is locked to a certain sync speed by NBNco. This is a physical limitation that cannot be exceeded, the technology in effect shapes the connection itself. While the line may have a higher "attainable" rate, the actual sync speed governs the maximum bitrate. If your line is capable of for example 100Mbps, but you are on a 25Mbps plan, the NBN equipment modifies the line condition so that the customer modem syncs at 25/5.
I suspect the main issue is that Enex (who have been fairly useless) are not tracking when test users change plan speeds and ISP's. People probably joined on 25/5 plans, and then upgraded to higher speed plans, thus corrupting the data. Enex don't track or care if you change ISP's or speed tiers(unlike samknows with the ACCC program). There are likely MANY test customers on different plans or providers since they joined.
It just makes the whole program unreliable with data that simply cannot be trusted to be accurate.