In this issue, we have three articles around the theme of "TM: I beg to differ" or, if you prefer, "alternative views on translation memory technology". While the benefits of "traditional" TMs have been clearly demonstrated in certain contexts, the number of TM licenses sold worldwide is minuscule compared to the number of translators. It is not entirely clear why this is case; at the very least, it means there is room for new ideas.

First, the mysterious Translator X argues that TM benefits are often less than advertised and often wind up in the pockets of the customer, the MLV, the SLV, the translation agency… in fact, everywhere but in the translator's pocket. Several very real scenarios are presented. He clearly demonstrates that a little knowledge is more dangerous than no knowledge and that customer expectations really need to be managed.

Second, Phillippe Mercier (Telelingua) presents a new architecture for managing TMs. Rather than centralize all TMs in one big repository, or loading them up in one huge database, this new approach is to leave them where they are (all over the world) and work with them remotely, be they TRADOS, SDL, STAR translation memories or even a machine translation server like SYSTRAN. Any language resource that can suggest a translation is potentially useful. A true Web application…

Third, to load the bases, we present an article from Tim Hunt on the failures and future of Translation Memory. This article was already published last New Year's Eve, so many of you may have missed it (and maybe other things as well). In any case, we reproduce it here for your convenience. Tim's article describes the fundamental objectives we need to achieve to produce truly useable software for translators.

Food for thought. Bon appétit!