Mar 11, 2014

Across: The Great Divide

A leading figure in the international translation technology world once remarked to me that when translation agency principals get together they gripe about SDL, but when tool vendors get together, Across is the subject of complaint. I can only imagine that to be because translation agencies are usually smart enough to stay away from Across altogether and know little of the horrors awaiting behind its virtual barbed wire. The philosophy and implementation of Across is like a virtual gulag for translators and data; adopt this ill-considered solution and let the software's developing perpetrator, Nero, fiddle while your business burns.

In the 1980s I worked at the research center of a major international enterprise and saw the terrible economic consequences of a proprietary laboratory information management system (LIMS) which led to the loss of years worth of data because the data could not be migrated after the software provider failed. So I was absolutely astounded several years ago to hear an Across representative at LocWorld in Berlin defend the companies "unique selling point" of incompatibility as a "security measure" that corporate clients appreciate. Not smart corporate clients with a future, certainly. Without even trying hard I can come up with half a dozen means of five-finger discounting the "language assets" stored on an Across server; what I can't do is suggest convenient ways for someone burdened with an Across server to work with the majority of linguists who refuse to have anything with what one agency owner described as "the only CAT tool that pretty much guarantees you under 2k words per day."

I looked at Across myself some time ago and was shocked by its appalling ergonomics, which complicate the work of translators to an extreme degree compared to popular and interoperable solutions like SDL Trados Studio, WordFast, OmegaT, memoQ and others. Across is a trap which offers its users no real advantage and a host of liabilities for data management and work planning. The "advantages" are entirely for the tool provider because of client lock-in and the inability of Across users to migrate easily to a better solution which allows cooperation with a wider spectrum of qualified service providers instead of merely those desperate enough to sacrifice themselves for the crusts to be had with this bottom-tier solution.

Although I'm known for my personal preference for memoQ for the kinds of work I do, I am familiar with quite a range of tools, and I can endorse any tool with reasonable ergonomics produced by a stable team with good support and a commitment to interoperable data standards. With a clear conscience I might support the use of WordFast, OmegaT, SDL Trados, Ontram, STAR Transit, memSource, memoQ and others depending on the particular needs of a situation, because I know that the client will not be locked in and will have viable options of work with qualified service providers who may have optimized workflows involving other tools. Unfortunately, this is not the case with Across.

A recommendation of Across by me would be a deeply hostile and unethical act, in principle a statement that I wish the client to be locked in to an inefficient, costly platform that will give the advantage to competitors with more flexible means of work. And fortunately, I cannot think of anyone deserving of such a harsh sentence as Across.

Creator Antony Stanley. This file is licensed under the Creative Commons Attribution-Share Alike 2.0 Generic license.
Photo: Antony Stanley. This file is
licensed under the Creative Commons
Attribution-Share Alike 2.0
Generic license.

[Update from the 2015 Jaba Partner Summit:
There may be important changes ahead. There may be cracks in this virtual Berlin Wall and perhaps a hope for the future for some of those now so cruelly shackled by the deluded "security" philosophyof Across. Time will tell, and I will be very pleased if action is taken to implement the words I heard yesterday from an Across representative.


  1. "So I was absolutely astounded several years ago to hear an Across representative at LocWorld in Berlin defend the companies "unique selling point" of incompatibility as a "security measure" that corporate clients appreciate."

    Following the same logic those same corporate clients would be even better served if their translators were made to translate using dip pens and lemon juice for ink, so that not even the translators would be able to see what they translated. Yes, a bit difficult to use the data that way, but one has to sacrifice something for the sake of security... in fact, do you think we could sell this idea to Across?

    1. I hear that's a feature planned for integration in Across 7.0. Unfortunately, the supply of lemons is limited in Germany, concentrated primarily in the assembly lines for Opel and Mercedes Benz. However, Mutti Merkel has announced plans to call in Mediterranean debts and exact tribute... uh, repayment I mean... in the form of idle youth (dragons gotta eat y'know) and citrus fruit, so we can all look forward to this innovation soon.

    2. O.K. While the endless ;-( 'rainy day reading' discussions about e.g. MQ v Sodrat Dustio still flourish on various forums I'm surprised that anybody even asks questions about Across in the chatosphere! I cannot remember ever having seen a positive posting about it and having read several bellows of alarm from respected peers who fell foul of it for a first and last time, I am convinced...

  2. Oh, the excuses some organizations come up with to justify their questionable practices and/or launder their motives never cease to amaze me.
    No matter how far removed from common sense something is, the explanation is always along the lines of: we do it for you, our customer/user, because you don't know what you want or need, and certainly don't understand the complexity of the big picture; We, on the other hand, do.
    Sometimes those explanations are accompanied by technobabble to give off the impressionist that they are supported on facts.
    Why people still use Across despite its well documented shortcomings, and why do translators keep working with agencies that use Across (a rhetorical question) is well beyond my understanding?

  3. On my part this is a real case of schadenfreude. I was recently asked by a direct client to switch to Across. The advantage to the client would have been that on the basis of an analysis in Across they would be able to tell how much of a given text had been previously translated and then get a discount from me. Trouble is I'm not the only translator feeding my translations into the memory. Nobody seemed to want to answer my question about garbage translations and would I still get paid if I changed them? Another issue is: how much MORE software am I supposed to spend my (unpaid) time getting my head around? So I politely refused. At the time I knew nothing of the "well-documented shortcomings" of Across because I've been a big fan of Wordfast for years and have never really bothered with looking at anything else. My poor client. But that's what happens when you get people with posh business degrees who aren't linguists to manage the process.

    1. Your poor client is right. I can tell you quite a number of positive things about translation environments I don't like but which have provided real benefits to the users. I find it deeply distressing that I cannot cite any real cases with Across. Instead I can tell about a senior manager at Honeywell who told me how one of their German divisions had been forced to dump Across for translation but kept using it to "manage" terminology because they had to get "something" out of all that investment (the guy apparently doesn't know the saying about throwing good money after bad). Or about how one of the most technically adept agency owners I know who can put SDL Trados Studio, Rainbow, OmegaT and other tools through amazing paces and who works with a bewildering array of development tools which merely frighten me, and who has nothing but the most depressing things to say about his experiences with an Across server. Some of the best and brightest technical guys I know in Germany have taken the challenge of trying jobs in Across and report nothing but failure and frustration. Of course these guys know many tools and can get by with almost anything probably... but that gives them certain expectations of performance that a completely ignorant person will not be burdened with. There are many reasons to choose my favorite tool, memoQ. There are great reasons to choose SDL Trados Studio in some cases. Surprising and compelling reasons to go for the combination of OmegaT and Rainbow. Even a few things to say for WordFast, though it's one of the most limited tools I can name which approach respectability. Déjà Vu has strong points still, though I don't think it has a future. But it might, there's no lock-in and considerable interoperability with other tools. memSource is developing in interesting ways and may be worth a look for many. I could go on and name half a dozen other options which would not be high on my list of recommendations but which would have positive, sometimes unique things to offer and would not be damaging. But Across is where I draw the line. Its "unique selling point" of platform lock-in is poison. Pure poison. Damaging. I would only recommend it to someone for whom I would actively wish harm. And even then I would feel guilty.

      I want my competitors on an even playing field where they can do their best and bring out the best in me. Where the tools don't get in the way of talent and hard work, or at least not too badly. But if they use Across, that gives me an unfair advantage. For them it's possibly not as good as taking a dull letter opener to a swordfight.

    2. I have just cracked up at the thought of taking "a dull letter opener to a swordfight"! Lovely. Must remember to use that one myself.

  4. Why any translation department would implement Across Server instead of GlobalSight is beyond me. There is plenty of support from companies like LocalizationSMART, TCO is low due to no license costs and as far as I can tell it does the same bloody thing. At the risk of sounding like a mouthpiece for its owners (who I work with) I can say that over the past couple of years, it has become one of the most implemented translation management systems in corporate translation departments. Translators who download jobs from it can use an array of CAT tools, including OmegaT (which also has a team version for translators who work in parallel).

    As someone said to me once, German companies like to work with German companies. I can't think of another explanation for its success, if 25 implementations last year can be called that.

    1. I've observed the general preference of German companies to work with German companies where possible, and if the products or services are comparable, it makes sense to do so. The same would apply to French companies, Italian companies, etc. There is a lot of added value in smooth communication with a service provider who speaks the same language you do and does it well. However, most of the German corporates I know aren't stupid, and I think that if a company realizes the extreme future risk entailed by investing in Across combined with the current impairment of access to good translation resources (i.e. most respectable translators won't touch Across after an initial experience, even if you put a gun to their heads), they'll look at any one of a the many good alternatives. As far as I'm concerned, any option a company can manage while maintaining interoperability with translators using a variety of working platforms is a good choice. As recent disasters with Lionbridge and others have shown, all the fine and finely tuned technology in the world does no good if you can only get less capable (or even incompetent) linguists to translate your texts.

  5. On the Across front... there might have been changes: I've been recently contacted by someone at Across -- they wanted to demo their product to us, seeing if we were interested. When I said that I had heard that their product would not be compatible with other tools -- that it would not be able, for example to use a tool like Xbench to QA work done in Across, they told me that Across outputs Xliff files, so that any tool like QA Distiller or Xbench could be used to QA translations done in Across, and that Across could work with files done by any other CAT tool that could output xliff files (i.e., most of them).
    Now, the devil is certainly in the details, of course... one would have how compatible Across xliff files really are. But maybe they are moving in the right direction.

    1. OK, I'll have to look into the XLIFF issue. As of Spring 2015 that was not the case, but in a meeting with Across reps in Porto it was said that some (unspecified and possibly optional) "opening" was planned, but the promised follow-up for that did not happen, and since the company and its products are of little real interest to me I didn't chase after the latest status.


Notice to spammers: your locations are being traced and fed to the recreational target list for my new line of chemical weapon drones :-)