We have already noticed that the SYSTRAN machine translation system is available both to officials outside the SdT and to the SdT staff themselves. One experiment in developing the use of SYSTRAN further is undertaken with the collaboration of the staff of the language help desk (AGL 3). The two areas concerned are the nuclear sector and regional development policy, where the D.G. concerned already uses SYSTRAN without going through the SdT. In practice, the D.G. helps to build special glossaries for these two areas. The help desk staff then receive texts and send them for SYSTRAN translation, do the post-editing and extend the vocabulary where necessary in collaboration with the D.G.
Translators in the language units also help with development of SYSTRAN. One such experiment concerns anti-dumping legislation. The translators pick up the original text through the computer network, ask for a SYSTRAN translation, then provide feedback to the SYSTRAN development team by sending a copy of the SYSTRAN raw translation, their own translation and notes. The feedback is not organised through standardized bug reports, and the translators are free to choose the problems they will concentrate on. The translators involved in this kind of work found it satisfying to see the results of their feedback appearing in the form of improved quality of the SYSTRAN output.
Despite all this, there were still some reservations expressed about the use of machine translation. The fact that SYSTRAN is used outside the SdT with little control is not an unmixed blessing in this respect. On occasion poor quality unedited raw machine machine translations have even been published in the Official Journal. In these cases, as in the case of poor quality translations done in the requesting services, it is the SdT who gets the blame in the eyes of the outside world, and this is, quite naturally, resented.