If you ask people to research 20 witnesses, you should expect them to produce a list of 10-15 they are ready to vote for (and 5-10 to avoid). That translates to 6-10 consensus witnesses under the current 1.5 vote-to-seat ratio (which I consider correct).
In other words, having 20 consensus witnesses is asking most people to research 40-60 candidates (unrealistic).
As someone who agrees with your premises, I would say the post actually calls for a decrease. Which in turn amplifies the point tarazkp made.