Scuds and Flies


Pie in the Sky

Although blockchain has been around for a decade now (maturing to some extent but not bursting,) the rhetoric has largely remained the same. In the same way as they have been saying for years, blockchain supporters still claim it has the potential to transform industries, making each one of them more transparent and efficient. To be true, every new technology is usually heralded with bombastic declarations and generic pie-in-the-sky promises. However, since little progress has been made, some initial enthusiast have started to fear that the hype might eventually be detrimental to blockchain’s potential.

A New War of Religion

Indeed, despite blockchain being baffling to most and greatly debated, its fans—or better, votaries—are firmly convinced of its intrinsic disruptive capacity and dismiss any diffidence as due only to ignorance and bias.

To those who are old enough to remember, the debate around blockchain will possibly remind the diatribes between Mac and DOS supporters in the early ‘90s that Umberto Eco had brilliantly described in a 1994 piece for his back-page column in a popular Italian weekly magazine, later cited by Stephen Fry to introduce his new column in The Guardian.

Jumping on the Blockchain Bandwagon

Anyway, despite controversies and criticism, blockchain initiatives proliferate with the few braves trying to benefit from Initial Coin Offerings (ICOs) as alternative funding methods.

Therefore, it should not be surprising that some blockchain enthusiasts would try to bring this technology into the realm of translation.

In fact, around the end of last year, a new ICO-funded, blockchain-driven startup was launched to be a “marketplace for multilingual communication,” Exfluency. The first round of fundraising through private sales of coins was scheduled to start before Summer this year. The ICO is scheduled to close by the end of 2019.

The standard-bearer of this new startup is its CEO Robert Etches, who is energetically spreading the word about the project the company he’s with is committed to.

The Use Case

Echoing the title of the book that has inspired him, he claims that “blockchain is a truth machine”. While this is mostly true for cryptocurrencies, where the problem of double-spending is crucial, and blockchain actually helped solve it, it also involves terrific overhead and collateral costs that make any other use case hardly convincing. In fact, blockchain requires a massive technical infrastructure—estimates of x times more technology spend than a standard relational database and y times slower performance. This means that there must be a strongly convincing use case for a translation blockchain to justify such a slow, heavy infrastructure.

As a matter of fact, though, many purported blockchain use cases (payments, voting, digital ID, etc.) amounts to little more than the willingness to add a distributed and encrypted ledger where one was not really needed. What if a decade after blockchain first burst on the scene, no one has adopted a decentralized ledger at scale because they don’t actually need or want to?

So far, no such use case has been presented, or at least plainly and clearly explained, to justify a substantial investment.

On the contrary, for the many reasons discussed in several other posts in this blog, using blockchain for translation is like “trying to swat flies with Scud missiles.” (Law & Order fans will possibly spot this line from episode 16 of the thirteenth season, Suicide Box, where Judge Joseph Flint scolds ADA Serena Southerlyn.)

And yet, as Exfluency’s initiator and chairman of the board Krzysztof Zdanowski noticed during a flamed and yet enlightening Twitter discussion, when investing in ICOs, the blockchain use case should be checked first.

The Problem, the Solution and the Pledges

Blockchain may be the solution but, what’s the problem? In other words, what problem, or problems, is blockchain is expected to solve, and why?

Exfluency’s website impeccably lists some of the problems—real or inflated—afflicting the translation industry that, hopefully, the would-be platform is supposed to solve:

  • There are not enough professional translators to match demand;
  • Security, confidentiality, and privacy guarantees are impossible to uphold;
  • Quality is unquantifiable;
  • Language platforms are closed and proprietary;
  • Translation is too expensive;
  • Free online options for machine translation are owned and run by gatekeepers who will leverage your data for their own gain.

Unfortunately, no indication can be found on the Exfluency website or elsewhere in any the many interviews, articles and presentations about how any of these problems are expected to be solved.

One would expect that, with no regulatory framework, legal protocol, or track record to abide by to issue an ICO, some explanatory document be issued to help investors make an informed decision. Actually, when close to issuing an ICO, companies are strongly recommended to accompany it with a white paper, and in fact Exfluency announced his early in September last year as forthcoming.

Unfortunately, no white paper is still available from Exfluency, and this makes the pledges listed on its website arguable:

  • Data is secure;
  • Translation will be significantly cheaper and delivered faster;
  • The Buyer is awarded a share of the token value generated by new linguistic assets created in the text enhancement phase and thus also becomes a member of the Community;
  • Language ceases to be perceived as a cost center;
  • More and better multilingual communication will drive growth for your organization;
  • The self-regulating Community will standardize quality at a high level;
  • The social impact will be immense—people will communicate better; endangered languages might even be revived;
  • Personal data remains the property of the individual;
  • Full respect for GDPR and the anonymization of data;
  • Linguistic data remains the property of the Community;
  • Companies or groups of members can create confidential Gated Communities, e.g. to enable supply chain collaboration on confidential projects.

Given the number and scope of problems and pledges, a white paper is expected to provide a detailed use case and answer a few basic questions on how these problems will be tackled and what are the solution envisaged for each of them, as well as what the time span will be.

Maybe, although far from being exhaustive, the following questions might help focus onto the complexity of the problems and the pledges:

  1. How will linguists be recruited and retained?
    This question refers to the first of the problems listed in the Exfluency website. NMT will help on the side of speed and volumes and, possibly, costs, not on the amount of qualified—and available—linguists.
  2. How will interoperability be pursued?
    This is anything but a trivial issue, since integration remains crucial in every IT project.
  3. How are security and privacy and IP going to be guaranteed at all levels?
    This is an issue that cannot be liquidated with some generic statement, as each aspect has different implications at every level.
  4. How is the investment be guaranteed?
    Buying tokens is an investment, as risky as investing in hedge funds or subordinated bonds, aggravated by the general legislation ambiguity, with no guarantee for the initiative to be successful.

Proof of Concept

The main question concerning the use of blockchain in translation is, where’s the innovation? The translation industry is in an extreme need for innovation and has always been looking for innovation. True, to date, it has just been seen in a company doing the same things faster and cheaper, sometimes even better. Therefore, a clear use case addressing the translation-specific problems that blockchain is expected to solve, and how, is crucial.

So far, there is seemingly nothing in Exfluency’s proposition, as it emerges from interviews, presentations and articles, that cannot be accomplished with a standard database: No change or innovation is prospected in the business model, while there is nothing in the vision exposed by Exfluency people that is not yet in translation.

The problem with hyping things as much as blockchain has been hyped is in people within the industry incorrectly thinking that blockchain is essential to be successful. As a matter of fact, blockchain is a very compelling technology, with many serious pros and cons, and thus it is crucial to be clear about where and when it is suitable.

The typical use case for using blockchain is with cryptocurrencies, and private cryptocurrencies like the Exfluency tokens are a good example of where blockchain could be used in the translation industry. However, if it is so, this would be a crypto business rather than a blockchain business. In this perspective, it is reasonable to assume that linguists, service providers, and other contributors would be paid in the company’s cryptocurrency as well, and a major issue could lie exactly in the uncertainty of vendors accepting private crypto as a form of payment. At the same time, the services Exfluency might develop should be purchased in fiat currency. Exfluency could then buy back and burn a percentage of the coins held by investors and vendors, which could theoretically make the remaining coins more valuable, provided they will be negotiable and exchangeable for fiat currency.

Anyway, the information available on the website and from the Twitter discussion cited above eventually led to drawing a few realistic hypotheses of what the Exfluency platform might be.

First, it will be basically a cryptocurrency exchange platform where blockchain will be used only for smart contracts. Exfluency will issue its own private tokens to be used for payments. They will be based on Ether, the Ethereum coin, while Ethereum will be used for the blockchain to handle smart contracts. As a matter of fact, blockchain will seemingly not be used for any elements of a typical translation workflow other than smart contracts and payments. However, it is not clear what exactly these smart contracts are expected to regulate, maybe the transactions pertaining to a translation memory repository in the cloud? Maybe, every time a single segment is leveraged, its owner is paid. In this case, though, the heaviness of blockchain would make very hard to justify micropayments. It would be, again, like “trying to swat flies with Scud missiles.”

Secondly, Exfluency will raise funds initially through private sales of coin, then via an ICO. Around 40% of the fiat currency collected in exchange for tokens will be spent for platform research and development, 38% for marketing and promotion, and 12% for operations. The remaining 10% will go to counselors, accountants, and advisors.

Third, the would-be platform will essentially be used to receive the uploads of language data from users—who would buy tokens—and host an NMT system. In fact, once validated and cleaned, the uploaded data will be used to build NMT engines.

Fourth, presumably future customers could buy these NMT services from the company for their own translation needs.

Fifth, no information is available about the mechanisms underlying the Exfluency marketplace, especially as to token-fiat conversion.

It’s unclear how Exfluency will recruit linguists to validate and clean up the uploaded data, and how much it will pay them, but they will be almost certainly be paid in tokens. Users uploading data will be paid in tokens too, while buyers of the NMT outputs will pay in fiat. In this respect, the promise of making translation significantly cheaper and faster is the hardest to keep. Translation prices have been plummeted in the last three decades, but very little has been made to reduce the overhead, and speed processes. And when a technological innovation has occurred, all the major industry players have monetized any savings coming from its implementation. Why should Exfluency be any different? Although this may seem mischievous and perhaps simplistic, logic would have it that the innovative idea would consist in further compressing the fees to linguists, even if this would make them even harder to find and hire, especially given the assumption that “there are not enough professional translators to match demand” (which this does not appear to solve).

Finally, in his interview to Renato Beninatto, CEO of the translation industry research firm Nimdzi and member of Exfluency’s Whitepaper Advisory Board, Robert Etches stated that “There will be no CAT tools!” However, these would still be necessary as well as APIs, platform integrations, etc., but no information is available in this respect too.

Incidentally, the largely common insubstantiality in dealing with the realities of language data raises many and serious doubts around the feasibility of a data-driven translation marketplace, as discussed in a recent post and many others before. Also, the IP issue will remain outstanding, especially since blockchain provides no mechanism to validate data ownership.

Problems to Solve

With no white paper still available that might offer more detailed information, there are only the Exfluency website and Robert Etches’s interviews, presentations and articles to rely on. It is not much, and it is hard from that to figure a consistent blockchain use case for translation, or a glimpse of real innovation.

If the assumptions made so far are right solid, Exfluency will still have a long road ahead even after v1.0 is released. Paradoxically, running the ICO would be the easiest part. Building the platform in each component, running it, adding integrations to major platforms, and collecting data will be much harder and take much longer. Hiring the right linguists will be even harder, especially given the crypto payment model. The hardest and longest part will be establishing a two-sided translation marketplace with the full capacity of handling projects, resolving issues, maintaining the right balance of buyers to suppliers, etc. Incidentally, no one so far has been successful in building a true two-sided translation marketplace, and it is really hard to see how blockchain and cryptocurrencies could change that.

Along with the financial aspects, time will be crucial. Blockchain is an immature technology that is largely being hyped for its implication with cryptocurrencies, but the blockchain arena is widening constantly with new developments that are increasingly detached from the initial association with cryptocurrencies. Uncertainty is mostly what has pushed many large potential blockchain users to step back at least temporarily.

However, it would be unfair to completely dismiss a company that hasn’t really even launched yet, but the hype around blockchain is making answering some simple, straightforward questions more and more urgent. Therefore, the best advice possible is to peer through the window, at least until all the questions raised so far are fully and clearly answered.

Final Remarks

Reading every day, everywhere, how blockchain will change our lives, revolutionizing the way we do things is frankly becoming annoying, as long as there is still no successful blockchain implementation in the wild applied where there is a real need for it. This has been a long time coming, maybe because it is a solution looking for a problem.

In this respect, Robert Etches provided a noticeably great piece of advice when saying that “People just have to stop being greedy and do a modicum of homework. If, for example, the project is backed by a 12-page whitepaper and a team of two, then treat it as the fake news it is.”