Massively Parallel Technologies offers to let an expert examine their Amdahl's Law proof under non-disclosure agreement. The readers of HPCwire will be asked to accept this expert's authority that the proof confirms MPT's claims.
Epistemologically, we “know” much based upon acceptance of authority: a teacher says so; the paper predicts rain; the radio announces traffic reports. Knowing by authority is crucial to knowing anything beyond our personal experience. Whether something we “know” is an accurate abstraction of reality or not corresponds to whether what we “know” is true or not. This presupposes that Truth exists, but acknowledges that what we “know” may be inaccurate.
Therefore, choosing authorities to be believed must be nuanced. For example, a news organization may have political bias in their selection and reporting of stories. Readers, aware of such bias, can then judge veracity of stories themselves. Ofttimes authorities will hide their biases and conflicts of opinion projecting themselves as paragons of impartiality. What one “knows” about the authority strongly determines whether one should believe what they say is proximate to Truth. Partiality is not necessarily bad, as long as its honestly acknowledged. Just because a real estate agent emphasizes a listing's best qualities, doesn't mean they're lying or it's not your dream home, but buyers should hire their own appraiser before making offers.
Proof supersedes authority. Proof asks for no acceptance of authority at all. Proof asserts the truth of its proposition with a sequence of theorems each derived by a sound inference rule from given axioms or theorems listed earlier in the sequence. Informal proofs may play fast-and-loose with these rules in the interest of brevity and comprehensibility, but it should be clear how a formal proof could be constructed from the informal proof, if need be.
Those that doubt the truth of the proposition may examine the proof for themselves. Having satisfied themselves that each of the theorems used a proper application of a sound inference rule, doubters can believe the proposition is True with highest certainty possible. The expertise, bias, and conflicts of interest of the authority offering proof are immaterial.
If MPT has proof, why do they stoop to smothered authority?
If MPT has patentable technology, fire-off provisional applications and come clean. Post the proof or give reason to suspect cause for reticence. This reader recommends HPCwire decline MPT's challenge and demand that MPT's proof be scrutinized by all.
Brian R. Larson
Chairman, Multitude Corporation
[email protected]
Comments to be considered for publication can be sent to HPCwire editor Tim Curns at [email protected].