Braun Tibor, Schubert András (szerk.): Szakértői bírálat (peer review) a tudományos kutatásban : Válogatott tanulmányok a téma szakirodalmából (A MTAK Informatikai És Tudományelemzési Sorozata 7., 1993)
MARTIN RUDERFER: The Fallacy of Peer Review: Judgement without Science and a Case History
174 RUDER I ER: T IIK I' A I.I ACY O l PEER REVIEW 4. DISCUSSION OF REFEREE REPORTS There were nine reviews of the paper (Q,S). The two referee reports (C,D) received by the author with the first rejection and the two (M,N) with the third rejection are now discussed. Reports of the other five reviews are not available for analysis except for verbal reports that one rated the paper as "excellent" (F,G) and that another was generally negative (I). Report C. This referee made two fundamental errors. In his first paragraph he referred to a published criticism' 2 3 > relevant only to the frequency discrepancy of the RGO clock at Greenwich Observatory which is specifically excluded from the analysis, as detailed in E, and which is irrelevant to the Cannon-Jensen finding based on the other six clocks. In the second paragraph, the "synchronization procedures" refer to the lack of absolute calibration of the six clocks to the international second. This affects only the Cannon-Jensen theory which requires absolute accuracy and not the finding itself which is an experimental result and hence exists apart from any theory. The adjustments to the atomic clocks by the time laboratories to slave them to the master UTC clock maintained by the Bureau International de 1'Heure is exactly what is required to test for the proposed explanation, as discussed (p. 402) and elaborated in E. This rejection is a graphic confirmation of the lack of scientific rigour in the judgement process. Report D. In his paragraph (a) the referee alludes to "imprecision and inconsistency" without proper justification. He erred in stating that the product of frequency and time in equations (6) (p.389) indicates that frequency is a constant. Constancy of phase, i.e., fT = fjT t, where f is frequency and T is one-way travel time between transmitter and receiver, is demanded by the Lorentz transformations, as cited just below equation (8) (p. 390). These only require f/f; = T/Tj. The frequency change in equation (11) (p. 390) is the well-known (classical) change due to phase modulation for a relatively moving observer, as stated. It is evaluated for an observation of a one-way propagation over a given duration T which therefore merely serves as a boundary condition. Because the referee may have been confused by the brevity of the original derivation it was expanded as noted in E. The paragraph added to the paper (p. 400) to quantitatively evaluate the referee's implication in (b) that the neglect of Sun's gravitational potential falsifies the paper showed this cause for rejection to be unfounded. In (c) the error of referee C is repeated for the same reason the CannonJensen theory was rejected — inadequate absolute clock accuracy — despite the discussion disclaiming the relevance of absolute accuracy in the rejected paper (pp. 401-2). The discussion in E and in the follow-up paper (p. 406) may be further clarified as follows: A systematic difference between two clock rates, as from any physical cause, results in an ever-increasing difference in clock readings which must eventually become sensible. However, random drift of clock rates over a long period results in essentially no difference in clock readings. Thus, if the magnitude of a randomly varying rate initially exceeds a systematic rate difference, the latter must still become observable over a sufficiently long period. Randomness of clock rate is maximised by the practice of the time laboratories to compare and adjust the individual UTCj clocks to the average (UTC) of a large number of free-running clocks. Any long term
