Simply, PI to network topological leaves doesn't scale.
On the contrary, PI *DOES* scale and it has scaled because there is no direct one-to-one connection between issuing a PI assignment and consuming a global routing table slot. Far more worrisome is the practice of issuing multiple non-aggregatable blocks to a single AS and the tendency of some ASes to use multiple global routing table slots for traffic engineering. Those things have greater impact on global routing table size than the number of new PI blocks. In addition, there is no consensus that the global routing table is becoming too big for providers to cope. If there was an imminent problem, then network operators would be meeting in public forums to define the scope of the problem and to agree on actions to mitigate the problem. If that were happening then RIPE could leverage the existence of this consortium of network operators to incorporate global routing table issues into its policies. But since this hypothetical consortium does not currently exist, RIPE can find nobody to take responsibility for global routing table impact and therefore cannot include global routing table size in its policies. --Michael Dillon