- ....
^{1}
- Recall
that $200 is the maximum possible value of
*E* to any client under
the TAC parameters.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

- ... so.
^{2}
- As
computed by Shou-de Lin of the TAC organizing team.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

- ...
goods.
^{3}
- The general allocation problem is NP-complete, as
it is equivalent to the set-packing problem [Garey JohnsonGarey Johnson1979].
Exhaustive search is computationally
intractable even with only 8 clients.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

- ... skyrocketed
^{4}
- With just 2
high-bidders, the only way to have the price escalate would be if they
bid for a combined total of 16 rooms of the same hotel type. That
could only happen if all of their clients were to stay in the same
hotel on the same night, a very unlikely scenario given the TAC
parameters.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

- ... average.
^{5}
- In general,
*ATTac-2000*'s average score
decreased with increasing numbers of high-bidders, as games became
more volatile.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

- ... earlier
^{6}
- This change has been adopted in the
specification of TAC-01.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.