Invited comments on “Centralization of document searching facilities”

tlllevah/ated [':'()Ill a rise viewpoint..\lthough work h::s c(>ntiuued it: l:ltille:'t)t[s i'esearch grOllt)S there ll:t,.,e t)ee:t It() draHtatie !)r(~:tl¢-throughs. .\ccordiugly, the statement di:ect]V preeediug the recontntendat[ons in our Fe[)(tFt OoBt inues tt) hold: Although prelimi:mry experime:::al results appear I~r(>mis[ug, ]arger-seale tests must be ettnducted, ancl a ntunber of te(}:::]e:ll l)r()ble[~:s nttlSt [)e solve(] before :: system [ue(wpor:::ing these cotieepts eau be developed which wit: be c,f stt[ticiet:t power" te enal.)lo high-perfcn'mauce searehi>g of very larg(centralized collections such as those considered here. In other words, while the promise of the associative searcl:ing methods has it: n(> way ciiminished, there is ~o reason today, auy more thaa there wns two years ago, it) p[au on using assoeiative methods, or any other machine te(:hnique, as a means f,o:' obtaining high quality exhaustive searching of very la:ge eo]lections. Nonetheless, long range planning does dietnte t}lat when the statistical associative techniques become available for pilot operation, as will be the case in at least one major gow, rmuent installatiou shortly, the associative system should also be sub-jeeted to rigorous quantitative evaluation of the type indicated iu reu, ommeudation 5 of our study. This will at le~tst elarify some of the scientific questions under dispute. In conclusion, we must emphasize the fae: that the foregoing The admitted shortcomings of the AI)L report oonstitute the best rebuttal for some of its eon01usions. The statement about the reeoramendatios: of burden of proof begs ~he question of Lord Coke, "If [ am wrong, what makes you right?" I believe, however, that the basic issue it: the ADL report seems to have bees lost in a cloud of argument about coordinate indexi~g and associative indexing, ete. In d:e content of what follows, the above argument is akin to an argument about the choice of paint for a new building. In order to bring the subject in its proper perspective, I should like to state what I believe the real basic issues are. We have presently hundreds of institutions of specialized information eenters supporting various research activities in government , private corporations and m:dversit.ies. Each of these specialized information centers is a treasure house of long cultivated spec, ialized skills for processing I:eehnical information. ]';aeh of these centers has acquired expertise in lhe subject: matter it covers, and knovns the best way to handle some of the specific problems the center may have. The existence ,of each one of these oenters is …