网络上社会系统不可避免的腐败-优快云博客

An interesting CNBC research article, “Amazon is filled with fake reviews and it’s getting harder to spot them”, exposes a long-standing problem: the sharp decline in the quality of Amazon reviews as a result of companies buying positive assessments for their products or negative ones for competing products on Facebook and other social networks.

CNBC的一篇有趣的文章“ 亚马逊上充斥着虚假评论，而且越来越难以发现它们 ”暴露了一个长期存在的问题：由于公司购买了对其产品或产品的正面评估，亚马逊评论的质量急剧下降。对Facebook和其他社交网络上竞争产品的负面评价。

Beyond the specific case of Amazon — which has recently suffered similar problems with Amazon’s Choice — the problem of the corruption of social systems on the web is, for me as a researcher, fascinating: as soon as a social-based metric, be it ratings, likes, favorites, followers or any other, acquires a certain popularity, schemes aimed at obtaining a profit by falsifying and distorting it automatically appear. As soon as a network or a scheme with a social base reaches a certain level of popularity, clandestine services destined to sell fake reviews proliferate on it, quickly finding interested parties willing to pay for them, and which end up, if no decisive action is taken, destroying the site’s value proposition.

除了Amazon的特定案例( 最近在Amazon的Choice中遭受了类似的问题 )之外，对于我作为研究人员来说，网络上社交系统的损坏也令人着迷：作为基于社交的指标，它立即成为评级，喜欢，喜欢的人，关注者或其他任何人，都获得了一定的知名度，旨在通过伪造和扭曲利润来获取利润的计划会自动出现。一旦具有社会基础的网络或计划达到一定程度的流行，原本打算出售假评论的秘密服务就会泛滥成灾，Swift找到感兴趣的人愿意为之付费，如果没有果断的行动，那最终会成为现实。，破坏了网站的价值主张。

This is a big issue for Amazon: in a crowded marketplace with many different options for every product, consumers tend to trust reviews. If those buyers start to believe that the reviews are simply bought and sold to the highest bidder, their trust in the site could crumble.

对于亚马逊来说，这是一个大问题：在拥挤的市场中，每种产品都有许多不同的选择，消费者倾向于信任评论。如果这些购买者开始相信评论是简单地买卖给最高出价者，那么他们对网站的信任就会瓦解。

Will Amazon do anything to tackle the problem? Detecting user review patterns is easy, as is using browser fingerprinting techniques to flag when a user is managing several different accounts. A clear message needs to be sent out that these types of activities will not be tolerated, which means banning wrongdoers from the site, in addition to reporting and pursuing the services that offer them — which is more difficult, but at least signals the company’s commitment to increase the barriers to such fraudulent activities.

亚马逊会做任何事情来解决这个问题吗？检测用户评论模式很容易，就像使用浏览器指纹技术来标记用户何时管理多个不同帐户一样。需要发出明确的信息，禁止进行此类活动，这意味着除了报告和追求提供违规行为的服务外，还要禁止网站上的不法行为-这更加困难，但至少表明了公司的承诺增加此类欺诈活动的壁垒。

The problem is usually the result of growth hacking schemes by the companies involved, which tend to be extremely short-term: a more active review market is usually seen as a positive indicator, which in many cases leads to increased popularity, activity, time spent on the site or sales. In practice, few companies tend to consider the problem until it is too late and they find themselves in the headlines and with their credibility undermined.

问题通常是所涉及公司实施的骇客入侵计划的结果，这往往是非常短期的：更活跃的评论市场通常被视为积极的指标，在许多情况下，这会导致人气，活动和花费的时间增加在网站或销售上。实际上，很少有公司倾向于考虑该问题，直到为时已晚，他们发现自己已经成为头条新闻，并且信誉受到损害。

Another obvious option is machine learning: the usage patterns of people taking part in these types of schemes are usually easy to recognize algorithmically, because their activity is restricted to these reviews. Again, Amazon is in a position to observe from on high the entirety of activity on its ecosystem, however vast and complex it might be, and could easily take action to prevent misuse.

另一个明显的选择是机器学习：参加这些类型的计划的人的使用模式通常很容易通过算法识别，因为他们的活动仅限于这些评论。再次，亚马逊可以从其生态系统的高处观察整个活动，无论活动多么庞大和复杂，都可以轻松地采取行动以防止滥用。

The pattern, in any case, is evident, and can be stated as a law:

在任何情况下，这种模式都是显而易见的，可以说是一条法律：

“Any system with a social base will experience, from a certain level of popularity, a corruption of its operations that will tend to destroy the value of the metrics used in it.“

“任何具有社会基础的系统都将从某种程度的普及中经历其操作的破坏，这将倾向于破坏其中使用的指标的价值。”

From this point on, we can only judge companies that experience this type of corruption in terms of their commitment to respond rapidly and seriously to combat it. Many years on from scandals such as buying followers or Likes on Twitter, Facebook, Instagram or, more recently, TikTok, these kinds of activities are still generally available, despite attempts to stop using them as a fundamental metric, and there is talk that these types of systems may be responsible for the creation of an ‘internet of lies’. But for the social networks involved, the problem is relative: their users are mainly interested in connecting with other people or to access information, and the reliability of metrics is usually a relatively secondary issue.

从这一刻起，我们只能根据承诺做出Swift，认真地应对的腐败来判断经历此类腐败的公司。自从在Twitter，Facebook，Instagram或最近的TikTok上购买追随者或Likes之类的丑闻以来已有很多年了，尽管试图停止将其用作基本指标，但这类活动仍然普遍存在，并且有传言说这些类型的系统可能负责创建“ 谎言网络 ”。但是对于所涉及的社交网络，问题是相对的：他们的用户主要对与他人联系或访问信息感兴趣，而度量的可靠性通常是一个相对次要的问题。

In the case of Amazon and other ecommerce sites, this is a bigger issue that affects the sales of products of companies that, ultimately, may even be forced to participate in such practices if they want to remain competitive, with the consequent direct economic damage. For consumers, the consequences of a fake review are sometimes greater, and can destroy confidence in the site as a whole. When a large percentage of your users or potential users know that your product evaluations are a big, fat lie, your value proposition and the trust that your users deposit in you and your operations as an online store can be significantly eroded.

对于亚马逊和其他电子商务网站而言，这是一个更大的问题，影响到公司的产品销售，如果公司想要保持竞争力，最终甚至可能被迫参与此类做法，从而直接造成经济损失。对于消费者而言，虚假评论的后果有时更大，并且可能破坏整个网站的信心。当很大一部分用户或潜在用户知道您的产品评估是一个重大的谎言时，您的价值主张以及用户对您的存款以及对您作为在线商店的运营的信任可能会大大削弱。

In any event, the problem has been identified, and we also know whose turn it is to make a move. If it is still easy to make a fake review on Amazon in the future, the damage to the company will likely end up being greater than that experienced by Facebook, Twitter, Instagram or TikTok. The time has come for Amazon to take action.

无论如何，问题都已被查明，我们也知道该采取行动了。如果将来仍然很容易在亚马逊上进行虚假评论，对公司的损害可能最终会大于Facebook，Twitter，Instagram或TikTok所遭受的损害。亚马逊采取行动的时候到了。

This article was previously published on Forbes.

本文先前已在《福布斯》上发表。

(En español, aquí)

( Enespañol ， aquí )