It's becoming clear that AGI is better seen as a spectrum of capabilities - see Google's underappreciated taxonomy, recent comments by Greg Brockman and earlier worldbuilding work done with Martin Wagah. The last year or two have essentially zoomed in to the vernier callipers enough to see between 0 and 0.1

Is it productive to similarly start thinking about ASI Taxonomy? Presumably one may be guided here by Kardashev scale.

ie ,

What model of the world would an ASI need to have to be equivalent to a Type 2 Civilisation ?

How is this limited compared to a type 3?

Is there an information theoretic superintelligence ( a type Omega Minus), st it would be impossible to develop a model of the world more effective than an IT superintelligence?

Can we make any robust security asumptions in any of these scales?