[blml] Undelivered Mail Returned to Sender

[Eric Landau]

My previous attempt to send this message appears to have been  
rejected; this may be a repeat.

On Aug 30, 2007, at 5:42 PM, Sven Pran wrote:

>> On Behalf Of Tim West-Meads
>> In order for UI to *demonstrably* suggest anything there must be a
>> difference between the UI and the AI.   Generally speaking once we  
>> have
>> taken AI into consideration we arrive at a position like "the  
>> options are
>> X and Y and while X is riskier the upsides are greater".
>>
>> If the UI doesn't *alter* the risk/reward ratio of X vs Y it  
>> cannot be
>> said to demonstrably suggest either one.
>
> 1: Difference between the UI and the AI is in itself no  
> justification for
> disregarding the UI.

If there is a difference, the UI *must not* be "disregarded".

> 2: Say that the "probabilities" without UI for selecting X is 60%  
> and Y is
> 40% while after incorporating UI they are changed to 100% for X and  
> 0% for
> Y. TD and AC will still need to judge whether this changed  
> situation is
> enough to disallow a selection of X because of the UI. We will have to
> consider whether the 60% weight for X shall be sufficient or if the
> increased weight for X caused by the UI shall be considered  
> significant so
> that X no longer shall be an acceptable alternative.

There's no need for such a judgment.  That the odds changed at all  
means that new information was "made available", hence L16A applies,  
even if the difference was minute.

> 3: And finally you have formulated a correct statement: If the UI  
> doesn't
> alter the risk/reward ratio it cannot be said to demonstrably  
> suggest either
> one.
>
> I can agree with that

As I suggested in a previous post, I do not believe that Sven and I  
disagree.


Eric Landau
1107 Dale Drive
Silver Spring MD 20910
ehaa at starpower.net