Well, Peter has criticised the Ishihara plates on a number of occasions for being too strict and finding too many false positives.
At least a lantern test should be reasonably 'honest' in that it's attempting to test colour vision using a task that's as similar as possible to the real life task of seeing navigation lights at night.
According to the web, the CAA now uses a computer based test, which I'll have to learn about - I thought there were other options such as Nagel anomaloscopy and the Farnsworth-Munsell test, which have their own problems.