Quid autem vides festucam in oculo fratris tui et trabem in oculo tuo non vide
What about a little experiment, the kind kids run in science labs at school? Set up Google Alerts for “translation industry” and keep track of the alerts. You will find that the news you receive is almost exclusively about the machine translation market, and occasionally about the Asian market (apparently by far the only interesting market), except for some sporadic ‘review’ from the industry’s most industrious press agency.
This is a plastic demonstration of the irrelevance of the translation industry, which goes far beyond the much-complained disaster effect, the one according to which translation only becomes relevant when it is bad, and maybe generates worse disasters.
Fueling Ambitions, Teasing Vanity
And yet, it is astonishing how some eggheads still think they have the power—and the ability—to educate customers, especially the largest ones, the ones that rock their industries.
Foolish? Maybe. Preposterous? Definitely. Obnoxious? Sadly.
Simply put, those people should rather look at the industry they are apparently proud to work in and try and fix the many flaws and faults in it.
Let’s be honest, it is a lot easier to write a piece explaining what is wrong with someone else’s vision of your industry, thus fueling the ambitions and teasing the vanity of its players, than it is to make a considerable effort to correct its distortions. And this is even truer if the piece is written while knowing that only those in the same industry as the author are going to read it.
And yet, even though they remain isolated, from time-to-time voices rise to say that the irrelevance of the industry is due to the lack of a serious marketing—if not lobbying—effort of trade bodies to try and launch some translation branding. But people prefer to organize events where they can talk and talk (most often fruitlessly), because it is much easier and possibly equally profitable, at least in the short term. In fact, building a brand takes plenty of time and money and resources. This also explains the booming of online events, all equal, even more so than before, all aimed only at gaining visibility in an industry that burns it up like alcohol.
So, before pointing one’s own finger at the organizational structure of most companies as the root cause of their translation problems, one should, evangelically, look at why the translation industry and its players enjoy a modest reputation.
For more than fifty years now, many larger companies in virtually every industry have been ISO 9000 certified, meaning their modus operandi is process-based, while their organizational structures, even when hierarchical and not heterarchical, are more and more project-oriented. So, there is no dichotomy between processes and projects, as the first embed the latter, and the latter must comply with the former. The problem lays in seeing processes as monoliths, to leave as such as much as possible. This has been the major reason for the translation industry to develop and pursue its own vertical quality standard while all other industries have been having no problems in implementing ISO 9000s since the very beginning.
ISO 9000 (and its predecessors) are based exactly on the assumption that processes run in repeatable cycles, each of them thus being predictable and scalable. Repeatability, predictability, and scalability are the foundations of quality management.
When, almost twenty years ago, on the initiative of EUATC, the European Committee for Standardization summoned the BT/TF 138 working group to write a translation quality standard, the future EN 15038 (mother to ISO 17100), many suggested to focus on a guide to an industry-specific interpretation of ISO 9001.
The reasons to follow a different approach were twofold: On the one hand, the peculiar nature of translation and the related services and the alleged inadequacy of ISO 9000 for translation quality management; on the other—much more credible—the assumption that certification would be much too expensive and out of reach of most LSPs, which would decree the failure of the initiative.
So, it is really not appropriate to try and fleece other industries that are certainly more structured and have a well-defined brand, as well as being equally—if not more—prosperous, even without any specific process standard. Indeed, many other industries have agreed on overly complex standards, such as those for arms and ammunition, transport (from amphoras to pallets and containers), audio and video formats, software engineering models and even fries.
Quality, Metrics and Measures
Ça va sans dire, most industries have also developed their own quality metrics and use them consistently for their everyday measurements. And the results are there for all to see.
On the other hand, William Thomson, 1st Baron Kelvin, is known for having said, “When you can measure what you are speaking about, and express it in numbers, you know something about it, when you cannot express it in numbers, your knowledge is of a meager and unsatisfactory kind; it may be the beginning of knowledge, but you have scarcely, in your thoughts advanced to the stage of science”.
Translation quality has always been hard to explain to industry outsiders because they are typically accustomed to be held accountable on the base of measurable data and cannot see how the idea of translation quality as discussed in the translation ecosystem can be suitable to them. Indeed, it is mostly meaningless as well as useless to any inexperienced buyer of translation services, and sometimes even to knowleadgeable customers.
What if quality were pointless and there were no such thing as translation quality? In business, quality is a given, but it is also a relative concept, even—if not especially—in translation, as a recent study on AssessingHuman–Machine Parity in Language Translation reveals, with professional translators showing a significant preference for human translation, while non-expert raters did not. Guess who pays for translation?
Anyway, never is any measurement good per se. Indeed, wrong metrics can lead to unintended consequences, and when a feature (like quality) is picked as an indicator, it ceases to function as that, because people start playing it. Also, some key indicators, like innovation, are simply impossible to measure at all.
The idea behind processes running in repeatable cycles, each of them being predictable and scalable, makes process automation possible.
As TechCrunch reports, “a lot of trend lines [have been] moving throughout 2020 and into 2021 around automation.” Automation today encompasses workflow, robotic process automation (RPA), intelligent business process management suites (iBPMS), artificial intelligence, process mining, and the movement to low-code and no-code development platforms. TechCruch also reports Statista estimating that the process automation market could be worth $74 billion in 2021.
The translation industry seems far away from process automation, and this is just another good reason to stop trying to educate customers, especially in those areas where they are so much farther ahead of most translation vendors.
Undoubtedly, the potential of the process automation market is drawing the attention of investors and this might explain also the recent VC inflows into the translation technology region.
In this respect, Semir Mehadzic, in a LinkedIn post, recently raised a question that is worth some consideration. He wonders why not open up translation marketplaces (ProZ, Translatorscafe) and the TMSs that operate as such (Lokalise, MateCat, Smartcat, Text United) and combine their data since anyone of them can hardly collect many more (active) profiles than they already have. He argues that, beyond technological shortsightedness, the interoperability blocks that these platforms have built are just vendor lock-in.
This brings back to the reasons for the failure of the TAPICC initiative. With so many platforms for so few users, not only is a universal API an unrealistic effort, but it is also unforgivable sin of pride. It is indeed at least naive to expect any of these platforms to implement an API from a competitor, which, although free of charge, would never come with any warranty. Also, the choice of going for a full-fledged API instead of an OSI-like model have possibly undermined the chances of success of the TAPICC initiative since the beginning, as the reduced and hesitant involvement of LSPs to the project—especially to the more technical WGs—had immediately shown. Of course, this would mean—and actually had meant—leaving all the decision-making and maneuvering space to the industry’s bigwigs.
A major problem when choosing a full-fledged implementation over a model is the resulting complexity of any updates and integrations like, for example, adding search and hiring capability of translators, which is a basic step towards interoperability and disintermediation.
Finally, universal means “including or covering all or a whole collectively or distributively without limit or exception” (Merriam-Webster), and this envisions a single, rigid, immutable, fully defined monolithic process, so much for scalability, although saving repeatability and predictability.
Talking of interoperability, seventeen years after Tim O’Reilly introduced the concept of Web 2.0, interoperability is still nowhere to be found because service companies try and keep users on their own platforms instead of sending them somewhere else. Even if this means higher costs, at least in the short term, they always look at engagement as the way to long-term revenue growth. And this is true even outside the translation industry.
On the other hand, Shopify is the perfect example of exploiting the power of platforms. While Walmart has failed to compete with Amazon, Shopify may overtake it with nothing to sell directly. The 218 million people that have bought products from third-party merchants sitting on top of Shopify might not even know about its existence. Shopify is massively diversified, so successes and fails occur in the aggregate, and Shopify has been growing faster than Amazon by unlocking niche businesses that were never possible before.
When looking at how the translation industry sees digital transformation, one would inevitably come across blockchain and NFT (Non-Fungible Tokens), the hype of the moment, just like AI yesterday, BMSs the day before, and, going backwards, the API economy, augmentation, IoT, adaptive ML, etc. Name an hype and someone in the translation industry rode it, hardly implementing anything, though. Obviously.
At a closer look, one would find that there are always the same old people riding them, most often only knowing about the hype of the moment because they have read something on LinkedIn, or listened a podcast or, today, overheard a conversation on Clubhouse (the trending app), and reposting anything without paying much attention. To gain some visibility.
Maybe a little more humility would help.
In fact, it is absolutely not true that “There have been several initiatives to combine the translation industry with the Blockchain innovation,” there has been only one (suspicious) initiative. It is true, though, that, for the last four years, this has been mostly a subject for speculations.
The latest craze of NFT promises to revolutionize the art industry by creating unique verifiable pieces of digital content linked to the blockchain. As Shira Ovide notices, NFTs are neither miracles nor scams. They are an example of people rushing to judgment about basically anything new and novel and they will probably be no world-changing revolution. Probably it is not an entirely absurd bubble, either. As with any emerging technology, there might be a clever idea in there somewhere if we slow down and resist the hype. In fact, most things in life are neither glorious revolutions nor doom. And behind most novel ideas is often the possibility of something useful. The problem with hypes is that they always go along with greed, so any promise sounds inevitably as bullshit. This should make everyone sane run away screaming from anyone who claims they are the answer. Anil Dash, who helped devising NFTs, is furious about the hucksters who are inflating a bubble around them.
Also, even though NFT might provide a way to certify and authenticate ownership of pretty much anything, tangible or not, the issue of the energy they require remains untackled and unsolved.
Maybe we just wait for the as-a-service model to encompass blockchain—and NFTs—too. This would of course mean waiting for Amazon, IBM, and Microsoft to offer tools and platforms enabling businesses to leverage blockchain as a service. In a few years, though, these platforms will bring blockchain technology within reach of organizations that otherwise would have to face up-front investments and spend years planning and building infrastructure and developing skills.
When that time comes (and it will not come early), LSPs might be among the late adopters (as usual), assuming translation will still be a business by then.
A Realistic Future
Danish cartoonist Robert Storm Petersen is mis-remembered through a famous quote misattributed to Yogi Berra: “It is difficult to make predictions, especially about the future”. More recently, in The Visionary’s Handbook, Watts Wacker, Jim Taylor, and Howard Means wrote, “The closer your vision gets to a provable truth, the more likely you are simply describing the present”.
Therefore, it is easy to predict that, at least in the immediate future, an industry that is historically very conservative will be focusing on the same stuff as of now. Expect no sudden leap forward in innovation.
Also, to be crystal clear, blockchain is no innovation, even less so in the ways it is proposed i.e., to support a translation data marketplace, which is already unrealistic per se, or to create non-fungible tokens for certified translations, which would never justify, economically and morally, the disproportionate energy consumption that would hardly be reduced. Incidentally, the issue of certified translations could already be tackled and solved through a public key cryptographic infrastructure. The biggest problem would be in the changes for public administrations to introduce, but the same would happen with an NFT-based infrastructure. On the contrary, while digital signature is already widely tested, used and economically sustainable, NFT might present unsustainable costs and would certainly require much longer implementation times.
Incidentally, blockchain will in no way solve the problem of translation quality whatever blockchain hucksters say. Without providing any supporting arguments, obviously.
Therefore, for many reasons, the near future of translation is going to be more and more machine translation.
However, showing off is never good policy, because you can always be called upon to prove yourself. For example, in this respect, most LSPs have failed in implementing machine translation, especially in building their own systems, typically when not having the basic knowledge required. Custom machine translation is out of range for most LSPs that are used to saving pennies, like everyone in the localization world. As Kirti Vashee recently noticed in his interview to ProZ’s Paul Urwin, the salary component alone for the team at Google is probably $10 to 20 million a year, while computing resources consume another $20 million. So, successful machine translation ventures around the world today require about five to $10 million of investment.
Anyway, machine translation has entered the mainstream for a few years now, and more and more enterprise buyers expect it to be used. This goes along with an expectation for lower prices, extremely low in some cases, but most LSPs got caught unprepared and reacted and continue to act in the only way they know, by enduring the pressure and dumping it on freelancers.
The resort to machine translation along with the expectation for lower prices is the consequence of the ever-decreasing value of user documentation, which in turn is a consequence of the increasingly shortening of the shelf time of consumer products. Companies are less and less willing to spend money for stuff nobody cares for anymore. If they ever actually cared.
So, post-editing of machine translation (PEMT) is no new big thing, even less so as long as it is done using the same approach, models, and patterns as ever (as per ISO 18587). In fact, All PEMT compensations schemes have proved largely ineffective, mainly because of the legacy of the traditional translation quality assessment approach in human translators who are still called on for PEMT jobs.
Media localization is not going to be the next big thing either, despite the interesting new tools that have been released recently, as not much has changed in the way of working and running projects.
The COVID-19 pandemic has led many companies to speed up pace of technological innovation. However, the wave of digital initiatives must not be confused with business transformation as most of them aim at just staying in the game, rather than at developing and securing a competitive advantage.
In this respect, Zoom or Clubhouse are no innovation, especially as long as they are used for chit-chats and pointless or commercial presentations. They are rather the evidence of the inefficiency in software procurement, investment, and spending. Software costs seem to more and more add up to overhead costs, with technology increasing the latter rather than helping reduce them.
The COVID-19 pandemic has made networking be missed and longed, as if patting on each other’s back were the best way to overcome any hardship. When are events in other industries going to see people from the translation industry not just attending (although rarely), but contributing too?
The translation industry, just like any other industry, and maybe more, needs to re-invent its working models and re-imagine the resulting value. On the other hand, regardless of the number of digital initiatives implemented, no one can realistically expect to win by being the same as their competitors who are, incidentally, all doing the same things, although some of them at higher speeds.
Some now contend that the near future for the translation industry is in data. The tide is ebb so it may be true, many LSPs might be thinking of exploring this side of the business. Indeed, a few companies that have been working on it for a year or two, but they all outsource or even crowdsource data jobs because the talent gap is extremely large, and there is no sign that academia is going to fill it any time soon. Anyway, with compensations being outrageously low and most LSP badly lagging behind, who is supposed to run—and join—the upskilling?
Unfortunately, times are getting shorter and shorter and the strategies that might have worked yesterday, even the most conservative ones, are already obsolete today, and tomorrow’s will be obsolete at the end of the day.
Bob Dylan wrote The Times They Are a-Changin’ in 1963. To someone as old as yours truly, it may even look as if times have not really changed at all. Quite the contrary. And they will keep changing. When is the translation industry going to do the same?