The article demands payments into a fund as an " ... “assurance bond” large enough to cover the worst-case damages". The article goes on with an example - installation of blow-out preventer in the Mexican Gulf - to reduce the risk. Two thoughts:
1) While this example of the "preventer" would reduce the risk, it would NOT reduce the level of the worst-case damages.
We know this type of situation from nuclear debates: damages so high that it is impossible to insure against (but seem to call for bans to prevent alltogether), and damage frequencies so rare that we can't set reasonable annual insurance premia.
The move of President Obama to press the responsible company into paying a substantial sum into a fund even before legal procedures are set - this could be developed into a new formula for law. The aim of the article's idea, to put a price tag on 'normal' activities even before a large damage occurs, is valid. It's only the focus on the "worst-case damage" which will not easily work. Possibly, it could be possible to define a "highest expected damage", i.e. a risk-based assessment of possible damages.
2) We need to look at the interests being created. The Gulf tragedy was partly due to the non-implementation of the legal control measures. So, any new measure should at the same time be less corruption prone than now. A large fund, however, is extremely prone! I am sorry to say that a tax would probably be less so, as the money is fuelled into known channels. But I am aware that this opens new questions, too.