1 min read

Berlekamp on Scientific Discovery

Both of these quotes are from Key Papers in The Developments of Coding Theory, which Berlekamp edited.

On originality

The originality of a contribution is often difficult to evaluate. When a subject reaches a certain point in its development, certain problems become popular or "ripe" for solution, and then several authors may all find similar solutions independently. For example, the weight enumerator of Reed-Solomon codes was discovered in 1965-1966 by Assmus-Mattson-Turyn, Forney, and Kasami-Lin-Peterson. The MacWilliams identities for nonlinear codes were independently formulated in the fall of 1971 in Belgium, at Murray Hill, NJ, and in Southern California. In the latter case, all three sets of people were responding to the same stimulus, which was the paper of Kerdock, preprints of which had been widely circulated in the summer of 1971.

On discovery

The problem of trying to decide how to distribute credit among several independent authors is complicated by another scenario which has also occurred several times. In this story, Author A gets some exciting new results which he begins advertising in one way or another. Perhaps he gives some lectures, or circulates a preliminary version of his manuscript, or perhaps he even publishes a paper and sends out reprints. Author B then talks to someone who heard A's talk, or he glances at A's preprint or reprint, fails to understand it, but gets interested in the problem and begins working on it himself. Being a clever fellow, he solves the problem by a slightly different method, gets excited about it, and writes his own paper on the same problem. Unaware of A's real contribution, B makes only misleading reference to A's work, or maybe no reference at all. Eventually A reads B's paper. If he is flattered by the imitation, the whole story might yet have a happy ending, but if he is angered by the alleged plagiarism, a bitter controversy may ensue.