Highs and Lows of Translation Business: Interview with Don DePalma, CSA

technical translation - tecom

A giant toy airship with the logo of one of the exhibitors was floating across the Rhein-Main-Hallen in Wiesbaden where the tekom fair – the largest global event for technical communication – took place. Sometimes it looked like it was arbitrarily moving by itself, but sometimes you could see the man underneath who steered and guided the Zeppelin with a small remote control unit.

This “wagging the dog” provided the perfect backdrop for a series of questions that I was going to ask Don DePalma, the founder of Common Sense Advisory (CSA), “the preeminent market research firm in the language services sector”, according to the exhibition directory.

Can stock market analysts move stock prices? Do securities prices go up and down because rating agencies create a certain market sentiment? Can market observers maintain objectivity if the market responds to their observations and depends on their judgment?

Some months ago I asked Nataly Kelly, the co-author of “Found In Translation”, about a CSA statement on translation rates. It turned out that she was no longer with CSA, but luckily, Don DePalma kindly suggested to answer my questions and also meet for a discussion. I was very grateful for the opportunity to talk with the mastermind behind many survey and research reports, often quoted verbatim by large translation companies and no less often perceived as controversy by translation professionals all over the world.

It was a very interesting and very long conversation. It is certainly worth reproducing more faithfully and fully that I will do at some point later and elsewhere, probably for the BDÜ’s magazine. In the meantime, I can offer only an abridged version that would fit in a (still very long) blog post. (My blog is called Anmerkungen des Übersetzers, German for Translator’s Notes, so I also allowed myself a few comments, highlighted yellow.)

This conversation took place at the tecom/tcworld fair and conference on technical communication in Wiesbaden, Germany, on November 8th.

Mira Vozreniya, a funny Russian-sounding character in your popular book “Business Without Borders: A Strategic Guide to Global Marketing" made me think about your language background. Did you learn Russian or work as a translator?

Don: As a child of the Cold War, living within miles of a nuclear station, the West Point military academy and many Fortune 500 companies’ headquarters…

…the military-industrial complex…

Don: Right, I still remember the air raid drills. Once I asked myself why we were doing all this. It’s because of the Russians – the answer stuck in my mind. Later in college I chose Russian, the language of those people who wanted to destroy us, and also had an opportunity to go to the Soviet Union. My bachelor’s degree was in American and Russian literature, my master’s in Russian literature and language. I did a doctorate in Slavic linguistics. I did comparative analysis, generative grammar, computational linguistics, historical phonology… My dissertation was on the ablaut in Czech which is called přehláska. There were no jobs in Slavic linguistics at the time, so the option was technical documentation, and from there on technical marketing. I became an analyst, but still carrying my language gene. I was hired by Forrester [a technology and market research firm] because of my technology background, and because I knew more than most Americans about things outside the US at that time. In 1999, I originally started Common Sense as an e-commerce consultancy, but it was difficult to grow, we were bootstrapping. I went to work for Idiom [Idiom Technologies Inc., bought by SDL in 2008], and later on I left Idiom because of my book contract with Wiley [the publisher of “Business Without Borders” in 2004]. When I finished the book, I said why won’t we do consulting…

As a translator and interpreter, I worked for consultancies a lot. One of the German business trainers who I worked for is fond of telling his students about the American optimism and positive look at things. Americans, he used to say, always find a way to talk themselves out of a crisis. They keep telling you that things are going to get better until things start to look better and, finally, turn good. Your consultancy, CSA, is known to spread the word about the decline, telling how things are going down, at least in terms of translation rates. Why this un-American stance?

Don: Well, that is the data. We don’t make this stuff up. We collect the data, we normalize it, we clean it up, we sit down with the statisticians, we develop models, and if the data says this we cannot say that. We have to say what the data says. Otherwise, it would be irresponsible.

A colleague translator from Canada complained that the public Translation Bureau, the largest purchaser of translation in Canada, based business decisions on CSA and was driving prices down. Many fellow translators I know have the impression that Common Sense Advisory effectively caters to its target group, that is large translation companies, and helps them drive purchase prices down. As a business consultant, you should keep in mind the interests of your target group, shouldn’t you?

Don: We don’t cater to any group. The data comes in, we analyze it, we calculate it, we say importantly what it means. How a company, buyer or supplier, uses the data, it’s up to them. If somebody says I am going to use it to drive the market down, there is nothing we can do about that. But there is one thing that I am going to tell you.

One of our clients, a large translation buyer, was sitting down to do a tender. It turned out that several LSPs were Common Sense Advisory readers as well. They said it was the best negotiations they had ever had, because both sides had access to the same pool of data. This is a very important part of the practice. It is something that I have been saying for years. What we do is we try to help companies deal with information asymmetry. Any time there is a discussion among two individuals or groups, typically one side knows more about the issue at hand than the other.

Somebody like Donald Barabé, the former managing director of the Canadian Translation Bureau (CTB), now retired, sits down with all of the data that he has, and uses that to make his decisions. SAP does the same thing. They make their decisions based not only on our data, but they look at the economic data from the markets they are in, they look at the company’s earnings.

Here is what we know about ourselves and about the market. The best buyers are also ones that have specialized procurement teams, best in terms of getting the best deal for the buyer. These companies hire individuals who understand how to buy. Unfortunately, what happens in some of these cases, they don’t understand the translation business. They know how to buy rolled steel to make Škodas. They say I know how to buy landscaping or building services, air conditioning, electricity or whatever. Now I am going to apply the same buying. Unfortunately, this is not a good thing.

Ignacio Lopez, the purchasing manager at General Motors…

…who later went to Volkswagen…

Don: … and got arrested for stealing… In the States, he went to the GM suppliers and said, “Every year I want you to take 3 percent, 5 percent, out the cost of the part that we’re buying from you.” It works if you build efficiencies into the manufacturing, so that the cost of the part doesn’t go up. But with humans, you get to the point of diminishing returns. The big challenge across the market is, given the drop in translation rates, that it makes the practice of translation much less desirable than it used to be. With all this downward price pressure…

Well,allow me to disagree. Talking about tenders reminds me of this famous saying by the American astronaut, John Glenn: “As I hurtled through space, one thought kept crossing my mind – every part of this capsule was supplied by the lowest bidder.” Also, the data that we see in your reports are of such a macro character, it is like trying to determine the average temperature and blood pressure across the hospital. Does price really matter? Do you take into consideration that the markets are different – a bulk translation market is not that similar to the high-end translation market. Rose Newell, a colleague of mine, just wrote that “competing on price when you sell a service that requires intellect shows that you lack the intellect required to sell or indeed provide that service well.” My impression is that CSA data might apply to big players or purchase managers at large corporations. On the other hand, they are of not much relevance for top-notch freelance translators or boutique translation companies.

Don: I agree. It is much more nuanced. But we don’t only measure the average temperature, we do look at the morbidity rate as well, to continue the metaphor. We see companies that died because of the average temperature in the hospital.

But generally, you agree that there are several markets, you wrote that the whole industry is highly segmented…

Don: Highly fragmented, right…

The figures of professional associations like BDÜ in Germany show upward trends. I know many translators whose rates are going only higher…

Don: … in high quality, precision, professional translations. Those are the kind of translators who will succeed in the future. There will always be opportunities for better translators. Higher visibility services, like transcreation, very specialized marketing kinds of things. Customers ask for different levels of quality. One company that I talked to uses an airline metaphor: they’ve got Coach, Business and First. The larger companies realize that, depending on the application, they don’t need the same level of quality. For FAQ or some knowledge base, they may say that can be a little rough, but the availability is important. If it is marketing material, it has to be absolutely perfect.

Translation rates and quality


Once again, the airplane metaphor came in handy. I didn’t change my mind about various grades of translation ("Add garbage to diversify!") since I wrote that post, but other companies have other motivation.

Don: Then we also have the changing nature of translation. We found a high percentage of respondents who didn’t include all the features like T – E – P (translating – editing – proofreading) in their rates. Some companies provide translations with very little editing, that explains low prices.

So basically you agree that there is a divide…

Don: Oh, absolutely.

… and it would make sense to have a more differentiated look at the market, also subject-wise? Do you poll literary translators or include journalism translation in your surveys?

Don: We tried to do a breakdown, but it didn’t work. The longer the survey, the more dropouts we have. To get the level of detail that you are asking for would be to double the length of surveys. One of the next things I am going to include in our next surveys is the reseller factor [the percentage of services that LSPs provide to other LSPs].

Another question that I had when getting prepared for this interview: Do you think the principle of economy of scale can be applied to the translation business?

Don [laughs]: Well, this is a great question. For an individual translator there are certain economies from gaining large jobs. You get the capital to buy the tools, services, a bottle of wine to feel good, at the end of the day. For an LSP, it is important to understand that they are an aggregator of supply and demand. There are thousands of translators like you who can provide services to potential clients around the world, but – they don’t have the access to those clients around the world. It is not a problem for you personally, but a lot of translators are not that visible as you are. Agencies make them more visible not directly, but provide them an outlet to the world.

So an individual translator outsources such functions like marketing and sales to an agency.

Don: Precisely.

Do you, an expert in globalization, see disintermediation as a trend? After all, globalization also helps to eliminate intermediaries.

Don: Some companies are going directly to smaller suppliers. SAP is looking for smaller, single language vendors. There are companies like Computer Associates (CA) that go directly to freelancers, there are various other companies. But can you buy directly from the supplier eliminating all the risks and costs that may make you go to an LSP in the first place? If they go to you for a precision Russian translation, best in the field, and you decide to go on holiday, so you’re gone for two weeks? They have to ensure availability, the 24/7 coverage. Another issue might be security, that is everyone is working within the same framework. True, it is safer to work with a single supplier, but when everything is in the Cloud, and the translator doesn’t have any content on his or her machine… These are concerns that come up: availability and variable cost, it is probably the biggest issue. If multiple freelancers require management by the buyer of translation, for each individual that they engage with there is some percentage overhead. Somebody has to oversee this work. It is a calculus. That said, there is a change going on: professional services that were going to India or China ten years ago are coming back. It was found to be inefficient to send them to another time zone and also inefficient to send them outside the company if the company was losing intellectual value.

These are apparent advantages for a client to deal with a bigger translation company instead of with a freelancer directly: availability and volume. Turnaround time is getting ever more important, more than price. The biggest mistake that freelance translators make is to turn down jobs (being unavailable) and, generally, fail to increase their visibility for a potential, preferably direct, client (otherwise they have to deal with intermediaries).

Did you discover new trends at this exhibition? Are there any new fields that you are going to research?

Don: Our goal with Common Sense Advisory is to provide an independent, objective view of the market. We care about all the participants, but we don’t cater to any particular group. We sell our research to buyers, we sell to LSPs. It is not just quantitative data, it is also qualitative data, benchmarking and best practices. Project management, production models, marketing methods by service providers. On the buyer side, I just finished a report on machine translation. Another research will be focusing on software localisation in an Agile environment. The Cloud is critical, from the technology viewpoint.

Companies are getting away from plain old translation. They want to increase their value to the enterprise they serve. They don’t provide just translation. You do a better job than any of the vendors on the floor for Russian to German or German to Russian translations. You can provide better services. They realize there are hundreds of thousands of freelancers and 28,000 other language services providers. They cannot compete on translation alone, they need to show more value, to become strategic. A couple of weeks ago, Lionbridge announced a global email campaign management system. If you want to send out a thousand of emails to people in 40 countries in the right language, that is a service that previously might have been provided by an email house. But now Lionbridge says this is something that we can also do for our clients. If translation can be provided by so many companies, let us see what we can do on top and, if the client decides to cut cost, they cannot cut us, since we are critical. Be strategic – that is what everyone would like to be.

As a photography enthusiast, I was lucky to be a participant in a three-day workshop held by one of today’s most talented photographers, a month ago. He mentioned, en passant, that there is so much talk about the photography business, technology, software and devices, etc., but very little about photography itself. My observation is that there is so much talk about the language industry or translation technology, but very little about translation skills and quality translation. Don’t you think all this technology is a little bit overrated?

Don: That’s an excellent observation. At the end of the day, when you stop, v kontse kontsov, the whole thing is about what you are doing, that is communicating things to the reader. There are various levels of communication. There is clearly an artistic form. I hope that people like you who have a passion for languages and translation will never lose it, because it is important. But there is too much stuff to translate, that is the bottom line. So much stuff is created every day and never ever leaves the language in which it was created. Some day, machine translation might reach the point that somebody might be able to improve the output. Other companies don’t need machine translation at all, because they have something like over 90% leverage on their translation memory. They are building products that are similar, from version to version.

Also, there is an enormous number of languages into which no content is translated, because companies don’t have the budget. There is no money for it.

Another issue is that today’s university students don’t want to go into languages. I don’t know what it is like in Germany, but in the United States and in the UK, they are shutting down language programs. One thing is a perception that it is not important. The other thing is less interest on part of students. The big question is where is the next generation of translators who can do high-quality work going to come from? Instead of making cuts in university departments equally, across the entire university, like the sequester in the US, they cut modern languages. Kent State University that offers PhD in translation studies and other universities are constantly under pressure.

Did they look into CSA research to underpin their decisions?

Don: They looked into our research. But they say they don’t see the enrollments or they don’t see the upside to the investment as compared to sciences. In an export economy like Germany that would be really dramatic if the companies wouldn’t be able to get their materials translated…

To be true, I don’t think that an exhibition like tcworld could be very inspiring for those who want to study languages and become a translator. Big LSPs are striving for independence from individual language professionals. A perfect translator is an easily replaceable, eventually expendable “translation vendor”. As big Zepellins try to hover ever higher over the anonymous “crowd” ("There's big money in crowdsourced translation", wrote Nataly Kelly, when she still worked for CSA), the skies for quality translators are getting cloudier. The market is split between high and low. It is not much different in other sectors. But translators need to get more aware of the divide. They also need to raise the awareness among their clients.

Common Sense Advisory - Don DePalma

I am very thankful to Don DePalma for being extremely open and comfortable with my questions. It was fun to have such a conversation. At the end I asked how it feels like for an American company to have a three-letter acronym in its name (you don’t have to be an expert in historical phonology to get the idea) and be engaged in data collection. Don told me about an email he received from somebody who thought CSA might sound like a cover for another three-letter organization.

I think Common Sense Advisory does an important work. It is international, and that is what various national professional organizations still lack, although I seriously think that IAPTI could make up for it. Independent translators and interpreters need to have independent research, and CSA has definitely something to learn from. I, for one, have learned a lot from our conversation, so thank you very much once again, Don, for the opportunity to talk about today’s language industry, its highs and lows.


  1. Kevin Lossner (@GermanENTrans)’s avatar

    Interesting discussion, Valerij. I look forward to reading more of it. So what did DePalma have to say specifically about where his "dropping rates" data derive from? It is clearly not reflected substantiated by any of the survey data from the BDÜ, SFT or ITI with thousands of professional translators as respondents. This seems to be more of a matter of agencies trying to cut the throats of their competition and often cutting their own. The recent GALA discussion in Berlin on LSPs use of machine translation also shows the desperation of these losers to go where the companies they serve often do not even ask them to. With the oversimplified, poorly qualified data the CSA have on offer and the ways in which I have seen those data abused, I cannot see that there is any really useful contribution to market knowledge so far. Bottom feeders may apply a distorted context to the published data in order to apply pressure to a naive service provider, but most others will probably need the sort of data the CSA does not gather because the longer surveys would cause more respondents to leave the questionnaires incomplete.


    1. ValerijTomarenko’s avatar

      At the risk of oversimplifying, I can throw in my explanation. The number of low price jobs (and low quality translations) will go on increasing (content explosion, “so much stuff” to translate where there is no money, “acceptable” or fit-for-the purpose quality, crowdsourcing… Take MT into the equation and you have your picture). Prices for high quality jobs (specialist, precision, marketing etc. translations) are going in the opposite direction, but will never be able to offset the average trend caused by the growing share of low price jobs. It doesn’t make much sense to discuss averages (or does only if you want to influence your market or rather help your procurement). Are passenger car prices going up or down? How many cars will be produced and sold in India? It probably won’t affect BMW prices, but will influence the average. We should take care of our market and do our own research (that is the message). BTW a great playlist of GALA-MT “highlights”: http://goo.gl/hqaWPO Everything that you want to know about MT!


      1. Kevin Lossner (@GermanENTrans)’s avatar

        Well, there certainly are a lot of interesting revelations in the GALA clips of the playlist. Kirti Vashees’s latest post on PEMT also has some interesting figures, which make it clear that even in what a seller of those solutions considers to be an optimized MT system, the post-editing hamsters can't come close to the volumes a skilled translator can achieve with good quality in a relaxed day with a good voice recognition system.

        I would have to disagree with you about the importance of the CSA's work. They are simply part of a great, incestuous echo chamber of special commodity interests, some of them CSA staff who cite the significance and uniqueness of the flawed data gathered while studiously ignoring any contradictory information from more credible sources. But as I support your right to give your hard-earned money to a gypsy fortune teller if it pleases you to do so, I think that companies have every right to learn from their mistakes with a heavy investment in the recommendatations of the CSA and all the myriad pushers of machine pseudo-translation.


      2. Olle Gustafsson’s avatar

        TRANSLATION PRICES ARE FALLING – that has been the mantra of Common Sense Advisory since years. All the pseudo-scientific, unverified and unverifiable research climaxed in the statement that the average per-word rate for translation fell 34.71% from US$0.205 in 2010 to US$0.134 in 2012.

        Each time I come across this statement or its numerous reiterations usually by big translation companies I am reminded of the question mark above – did you know?!

        Did you know that the majority of translators, i.e. translation practitioners aka freelancers, were exluded from Commos Sense Advisory surveys?! Did you know that the data this statement is based on reflects the wishful thinking on the part of translation brokers?! Did you know or did you ask what kind of per-word rates this statement refers to (purchase prices, adjusted by the purchasing party)? Information asymmetry indeed.


        1. ValerijTomarenko’s avatar

          Yes, TRANSLATION PRICES ARE FALLING – that was the statement that I initially asked Commons Sense Advisory (first through Nataly Kelly) to comment on. It was exactly my understanding that freelancers were systematically excluded from CSA polls, so that any CSA statements were biased towards what translation providers – mind you, in the definition of CSA – led you to believe.

          The exclusion of freelancers from CSA polls and the refusal to consider freelancers to be “language service providers“ contrary to international standards is stated explicitly in “The Language Services Market: 2012”: “Our research has consistently found that only companies with two or more employees generate revenue significant enough to merit inclusion in our studies. So, our validation process included steps to make sure that the company was of sufficient size to be considered part of the general population of LSPs (Methodology, p. 41) and later on “we purposely remove one-person companies from our database” (p. 42).

          So when I received the feedback from DePalma (“I was the lead analyst on the global pricing project that you’re asking about. Published in September 2012, it was our fourth such pricing report since 2004. For the survey that underpins the research, we polled more than 3,700 suppliers in 114 countries. The suppliers included freelancers and small, medium, and large language service companies.”), I thought it was a mistake.

          However, when I asked Don if he was sure about freelancers, he showed me a spreadsheet and explained that freelancers were excluded from all the previous polls, but were covered in the 2012 report. So the 2012 report on translation pricing seems an exception.


        2. Kevin Lossner (@GermanENTrans)’s avatar

          Well, those figures from the CSA have always smelled a little funny, especially given that large surveys by three professional organizations (ITI, SFT and BDÜ) contradict these findings. For a while I assumed that the figures discussed might reflect increased competition among the bottomfeeders in the Linguistic Sausage Producer market, but even that doesn’t make sense. A decline on the order of magnitude the Common Sense Advisory claims would have resulted in more bankruptcies and consolidations than have actually occurred. I think the answers to these questions might be found in a careful review of what the actual composition of the groups on which these statistics are based is. If the CSA included data from freelancers in the 2012 statistics, this could of course totally skew the data and make any comparisons utterly invalid. And that would make the use of such comparisons quite unethical. So what is the actual breakdown of the groups included in the statistics for 2008, 2010 and 2012? Were the populations studied each year the same, or were there significant shifts which might affect the averages of rates reported?


        3. Olle Gustafsson’s avatar

          Well, Kevin, I don’t know what you mean by “smells funny“.

          I am not a translator, but I work for a translation company, and I had enough opportunity to examine CSA reports.

          Let’s talk about comparisons.

          The above cited “The Language Services Market: 2012” contains “Year-to-Year Change in Average Revenue per Employee by Country”. It would suffice to take a look only at a few European countries (beyond Europe the volatility is even higher).

          There is no plausible explanation for the jumps in revenue per employee in Belgium (+130%) vs. Netherlands (-25%) or in Germany (+100%) vs. Austria (0%) vs. Switzerland (-60%), there is no explanation for huge discrepancies in absolute figures for Germany, Austria (twice as much) and Switzerland (approximately 60% of the German number).

          The figures for Scandinavian countries are simply not true: you cannot have $250,000 in Sweden and $75,00 in Finland (speaking from experience). All in all, these are random data points that don’t show any trends. These comparisons won’t withstand a plausibility check. Statistically, they are irrelevant. They don’t make sense, and your assumption about different “populations studied each year” is probably true. In fact, it is the explanation provided by Common Sense Advisory: “The reason for a significant jump from one year to the next may simply be a change in the sample for that particular country.”

          It is their GUESS, and my guess is that the reason for the famous conclusion about DROPPING RATES (–34,71%) also “may simply be a change in the sample”.


        4. Piet S.’s avatar

          The saying about translators being literate, but not numerate, totally applies here. Neither I am a translator, but as a former scientist, working for a translation company, I started wondering long ago why nobody pays closer attention to the data published by Common Sense Advisory. The “mantra” of dropping rates and the conclusion that “the average per-word rate for translation fell 34.71% from US$0.205 in 2010 to US$0.134” results from a great oversight or error that IN VARIOUS YEARS, VARIOUS POOLS OF RESPONDENTS were polled. Imagine polling a group of 20% teenagers and 80% adults one year and a group of 80% teenagers and 20% adults next time.

          In 2010 the pool of respondents comprised 26,9% freelancers, in 2012 the percentage of freelancers polled amounted to 83,91% of the whole population. Since freelancers generally work for LSPs and have much lower rates (-30 to -60%), the average plunged proportionally (-34.71%).

          The appalling mistake is that Common Sense Advisory doesn’t take into account various categories of respondents and their percentages in the whole set. The averages are provided unweighted, across the whole datasets.



        5. Kevin Lossner (@GermanENTrans)’s avatar

          Piet, I think the rate differentials between the larger Linguistic Sausage Providers and the freelancers who often work for them is significantly more than 30-60%. I am aware of routine markups by some of at least 100%.

          In a data simulation I did using CSA population figures and an assumption of a 2:1 average ratio of agency to freelancer rates, I came up with a calculated difference in averages from 2010 to 2012 within 1% of the figure reported by the CSA. This was with the rates for agencies and freelancers remaining exactly the same.

          There are of course two possibilities for this flawed analysis: deliberate deception or pure incompetence. Regardless of what is truly behind the false report from the CSA, it serves the purpose of special interests who need such numbers as an argument to push whatever crowdsourcing or machine pseudo-trashlation snake oil they are selling or increase their margins by pleading "crisis" to gullible digital sharecroppers. I figure if these phony figures from the CSA didn't exist, companies like Smartling or Asia Online or thebigword might have to invent them.


        6. Shai’s avatar

          I'm jumping a little late here, and I don't know enough about the CSA methodology to comment about it, so I will just make a general comment.

          I have a mathematician, trained as a scientist, friend who is particularly impatient with what he refers to as "Generic, superficial, pseudo-statistical analyses for social-engineering purposes".

          He argues that so many surveys are ordered and conducted so superficially, sometimes maliciously and other times just out of lack in knowledge, that their results are often almost irrelevant, at best. He argues that bias in any research is an intrinsic aspect that should be tackled very carefully in the research design and data analysis stages, and even then it is still a factor. These surveys are superficial because they exploit the human weakness of many readers to get the bottom line quickly without bothering themselves to derive their on conclusion or carefully examine the results. Without quality information about the methodology, design, data quality, and the context of the research and the data analysis, the results have no relevancy to a particular case and context, and many marketplaces are so fragmented and segmented that an attempt to reach a one-size-fits-all conclusion is not very serious, if to be polite.

          “We collect the data, we normalize it, we clean it up, we sit down with the statisticians, we develop models, and if the data says this we cannot say that.”

          Is exactly the attitude he is impatient with and criticize: the portrayal of the process ultimately determining the data accuracy and quality as some technical, mostly boring and irrelevant, mechanical process, that steers the unsuspecting readers to jump even quicker to the "conclusion", in an attempt that they will cite it and spread the word, trusting that if you recount a disinformation enought times, it eventually becomes the truth. It is so easy to manipulate the results – even a simple yes or no question that is allegedly the simplest of all can produce widely different results depending on the survey population on how the question was phrased.


          He often like to use the following metaphor (which of course is oversimplified for the purpose of emphasizing the message): If you put one feet in a bucket of boiling water and the other one in a bucket filled with ice-cold water, statistically you are fine, in practice, however, you lost both feet; that’s statistics for you.



Your email address will not be published. Required fields are marked *