I believe it was on NPR on Friday, but someone (remote) was talking about why one would want multiple witnesses to testify. His contention was that with multiple testimonies, the "accurate" parts will tend to reinforce each other across testimonies while the "inaccurate" portions would tend to cancel each other out.
Suppose for a moment, that that is true. Could we use a web of testimonies to fabricate a semantic hologram? I think, to really pull it off, we also have to suppose that the "truthness" of a proposition will modulate in the juror's head. If that were the case, it should be possible to construct a web of testimonies (carefully orchestrated in time) which makes no specific claims on a particular proposition. This web could be used to create a peak of "truthness" for this particular proposition in the juror's mind at just the time the attorney calls upon the jury to consider the proposition.