At the risk of introducing yet more non-certainty, if not true 
uncertainty, I will add two further comments (and now it looks like 
Cantillon may well have been ahead of all of them, which would not be 
the first time).

Regarding Keynes and Knight on entrepreneurship, which Cantillon and 
others tie to "uncertainty," however defined, I would say that Keynes 
leaves many parts of it out.  He does not discuss such matters as 
organizing and founding a firm, hiring workers and motivating them, 
or how one gets a new idea for a product or a process, much less the 
details of how one raises financing, although he was himself involved 
in such matters in practice.  Where his view of uncertainty links 
with Knight and entrepreneurship is in his theory of animal spirits, 
which many Austrians do not like for reasons that remain mysterious 
to me.  In the face of non-measurable non-certainty, whether it is 
for epistemological reasons for risk or true ontological uncertainty 
(no probability distribution to find), the entrepreneur must 
ultimately make a decision to act, and at that point something beyond 
rational logic must be involved, which Keynes invokes as "animal 
spirits," although others will simply label "willingness to bear 
risk," which does not quite capture it, since "risk" is supposedly 
measurable, whereas uncertainty (or not-easily measured forms of 
risk) are not.  Again, while many think about Keynes's use of the 
term in terms of financial markets and speculation, he also clearly 
uses it in relation to real capital investments, and making those 
needed to start an enterprise are a crucial aspect of entrepreneurship.

Regarding the the matter of Bayesianism versus classical frequentism, 
there is a curious overlapping of the views of Keynes with some 
Austrians.  In his later periods Keynes criticized Tinbergeniann 
econometrics partly on grounds of the limits of classical 
frequentism, as well the existence of phenomena for which there may 
exist no probability distribution to discover at all (true 
uncertainty).  One of the limits is that for much of economic reality 
it is difficult, if not impossible, to carry out repeated 
"experiments" to have that wonderful asymptotic convergence to the 
"true" underlying probability distribution, as may be done if one is 
flipping a fair coin.  Most practicing econometricians simply assume 
underlying distributions to hold and proceed accordingly, applying 
measures and theorems without really knowing that they hold (such 
matters have been more on many peoples' minds since the recent events 
in financial markets have reminded us about fat tails and black swans 
and so on).

One finds a similar critique in Ludwig von Mises (not Richard), who 
was the mentor of that great student of entrepreneurship, Israel 
Kirzner.  At one point in Human Action, von Mises criticizes 
political pundits who post odds about the outcomes of forthcoming 
elections, reasonably pointing out that each election is different 
and that therefore there is no basis for invoking classical 
frequentism in the assigning of such odds (not precisely his 
language, but that was the point).  He clearly saw this extending 
well beyond forecasting electoral outcomes.

BTW, regarding those fair coins, even there one can encounter a 
difficult epistemological problem.  If we start tossing a coin, how 
do we know that it is "fair"?  Sure, we can physically examine it for 
imbalances or filing or whatever, but what if it starts simply 
showing one outcome all the time?  This is the case in the fabled 
philosophical discussion of probability theory that occurs in the 
opening sequence of Tom Stoppard's breakout play, Rosencrantz and 
Guildenstern Are Dead, when one of them starts tossing a coin that 
keeps coming up the same way (forget, but it might have been 
tails).  No way to know for sure.

Barkley Rosser