Let’s go back to 1983. Stanislav Petrov was a Soviet watch officer. An alert went off on his console – according to his monitoring system, there was an incoming nuclear attack from the United States. But rather than alert his superiors, who would have immediately ordered a cataclysmic counter-attack, Petrov considered the context of the information available to him. His training taught him that any nuclear attack would likely be all-out war, with dozens of missiles and support from every branch of the U.S. military. Yet there were no reports of anything happening, other than this handful of missiles. Petrov decided not to take action on the alert. And he made the right call. It turned out that the early warning system had mistaken sunlight reflecting off clouds for missile launches. By questioning the validity of data, Petrov averted nuclear war.
让我们把时间拉回到1983年。斯坦尼斯拉夫·彼得罗夫是一名苏联的一名值勤军官。突然，控制台响起了警报。监测预警系统显示，美国的核攻击即将到来。但是，彼得罗夫对手头的信息进行了一番斟酌，决定不向上级汇报。因为一旦汇报，他们就会立即发动毁灭性的核反击。他接受过专业培训，知道任何核攻击都会导致一场全面战争，随之而来的还有数十枚的导弹袭击，美国军队各部门也会同时参战。然而，除了这几枚导弹警报外，监测预警系统并没有发现其他异样。彼得罗夫决定对警报不予理睬。而他此次的决定是正确的。事实证明，是苏联早期的预警系统将云层反射的太阳光误认为是导弹发射。彼得罗夫质疑了数据的合理性，从而避免了美苏核大战的爆发。
Companies today aren’t saving us from nuclear war, but they do face a major set of new risks from data-based decision-making. And in most cases, companies fail to grade truth in data.
如今，世界各大公司并没有将我们从核战争中解救出来，但它们却面临着基于数据决策所带来的一系列新风险。而且在大多数情况下，这些公司尚无法对数据的真实性进行分类。
If I were looking to do harm to an oil company, I could do much more damage by hacking your view of the world – getting you to purchase the rights to drill in the wrong places – than I would by hacking one of your oil rigs and shutting it down for a few days. Think about how many decisions are made based on data, and how vulnerable that data is to manipulation when no one is thinking about the information itself as a risk.
假设我想坑一家石油公司，我可以黑掉其数据系统——让其在无油可采的地方购买开采权——这比黑掉它的某个石油钻井平台，并让其关停数日所带来的损失更大。想想有多少决策是基于数据作出的，而当无人视其为风险时，这些数据又是多么易受外界操控。
This is the power of data – and the risk of not having strong data veracity.
数据的力量巨大，可失去了准确性，数据的风险也同样巨大。
Siemens is helping oil and gas companies address this threat today. They're partnering with cybersecurity vendors to embed intelligence into the sensors on board industrial equipment. The sensors take an aggregate look at the data fed to them to develop a baseline for normal behavior. With this baseline in hand, oil and gas companies can compare deviations and have a better chance of detecting threats and intrusions into their systems.
现在，西门子正在帮助石油和天然气公司应对这样的威胁。它们与网络安全供应商合作，将智能芯片嵌入工业设备主板的传感器之中。这些传感器可以对输入的数据进行综合分析，为正常行为制定基线。有了这一基线，石油和天然气公司就能够对比数据偏差，从而更好地检测系统中的各种威胁和入侵行为。
But it’s important to realize that the risks from bad data aren’t limited to intentional manipulation. United Airlines realized that inaccurate data was the culprit behind a billion dollars a year in missed revenue. It turned out their seating demand forecasts were based on decades-old assumptions about flying habits. This drove inaccuracies in pricing models. That data was true at one time, but it doesn’t make sense to base decisions about today’s airline customers based on data from decades ago. If your business has data-driven revenue, can you count on that revenue if you can’t trust your data?
但重要的是要认识到，错误数据所带来的风险并不完全是人为操纵的。美国联合航空发现，不准确数据是导致其每年损失高达10亿美元收益的罪魁祸首。原来，它对飞机订座需求的预测沿用了几十年前的乘客飞行习惯数据。这就导致了其定价模型失准。这一数据以前是准确的，但时至今日仍用几十年前的数据来确定航空公司乘客数量的做法显然是不合理的。假如你的企业收入是数据驱动的，现在数据没指望了，收入还有指望吗？
Every business should also consider the ways their own processes are incentivizing stakeholders to game their data-driven systems. Researchers in the UK have been studying how rideshare drivers game pricing algorithms – they orchestrate mass sign-offs, making the system think there’s a shortage of drivers. In order to get more drivers on the road, the system then increases prices. In a sense, the rideshare company’s algorithm is incentivizing its own drivers to game its systems.
每家企业都要考虑采取什么措施去激励股东与数据驱动系统进行博弈。英国科研人员一直在研究共享汽车司机是如何与定价算法进行博弈的——他们编造了大量的签名，使得系统认为出现了司机短缺。为了让更多的司机上路工作，系统就提高了价格。从某种意义上说，共享汽车公司的这种算法是在激励自己的司机与系统进行博弈。
Amazon provides a good example for how incentivize the truth: when they realized sellers were publishing fake reviews, Amazon revamped their business processes by creating a “verified reviewer” program, where reviews from those who had purchased an item were given more weight and appeared higher in the reviews section. Amazon also took steps to formalize a system that offered merchandise to consumers in exchange for a review, putting stronger controls over it and providing more transparency to consumers. That helps build trust in the Amazon brand.
亚马逊在如何激励真实性方面树立了一个好榜样：亚马逊曾发现卖家发表了许多虚假评论，随之改进了业务流程，创建了“经过验证的评论者”程序，赋予商品购买者更高的评论权重，并让评论显示在评论区更靠前的位置。亚马逊还曾采取措施，建立了向消费者提供商品以换取评论的系统，从而强化了对评论的管控，为消费者提供了更多的透明度。这有助于建立消费者对亚马逊的品牌信任。
You can build a practice to grade the truth in your data based on skills and capabilities you already have. Pulling from cybersecurity, data science, and analytics, you can build what we’re calling a “data intelligence practice,” and attack this new type of vulnerability throughout your business. By building a culture that emphasizes the importance of data veracity, you’ll also build confidence in your own data-driven decisions. And, critically, you’ll make yourself a trustworthy partner for people whose lives, jobs, and organizations are affected them.
你可以依托现有的技术和能力对数据的真实性进行分类操作。借助网络安全、数据科学和分析学，你可以建立所谓的“数据情报操作”，并在整个企业中攻击这种新型安全隐患。通过打造一种重视数据真实性的企业文化，你在数据驱动决策时将会更自信。最为重要的是，你将成为人们值得信赖的合作伙伴，因为他们的生活、工作和事业都将因你而变化。
（2021年安徽省第三届翻译大赛英译汉初赛原文）