1、 外文翻译 Search engines and the public use of reason Material Source: Ethics and Information Technology, 2008(9) Author: Dag Elgesem Search engines have become the central entry points for access to information on the web. Google and the other commercial search engines are thus increasingly powerful pl
2、ayers in the new information ecology. My focus here is on the power of search engines in relation to their users. There is an imbalance in power because the criteria that the search engine uses to rank its results are not transparent to the user. The question I want to discuss in this paper is which
3、 moral obligation if any this power puts on the search engine companies. It might be suggested that the search engines have no moral obligations to their users at all because they are private companies, offering a service free of charge, and that the user can switch to a different search engine with
4、 no cost. But this cannot be right, I think. Millions of people use search engines every day to search for information. Consider the question of paid hits in search engines. Google and most other search engines maintain a clear separation between so called organic results and paid hits, i.e. ads. On
5、 the Google result page, for example, the organic hits are listed in the middle of the page and the paid hits in a column on the right side. Sometimes there are ads also on the top of the page. This is a helpful way to divide up the results from the users perspective. But I dont think Google or the
6、other search engine companies are morally required to separate their results in this way. In the late 1990s there the company Go To launched a search engine based completely on only paid hits: the best ranks on the result page were awarded to the highest bid in an auction. Go To failed to attract ma
7、ny users and must be considered a failure as a search service. This is not surprising because a search engine based on paid hits cannot be trusted to return the results that are the most relevant to the users query. However, I cannot see that there is anything morally wrong with Go Tos strategy as l
8、ong as the users were informed about the basis for the ranking. It was a great model for selling ads on the web, however, and Go To was thus later acquired by Google who now uses their auction model to sell the ads that appear on the right hand side of Googles first page of results. But at the same
9、time Google learned from the failure of Go To that users do not want a search engine where the organic results and the paid hits are mixed. Hence, Google chose as their strategy to have a clear separation between these different kinds of results. The design of Googles pages is based on this separati
10、on and this transparency is probably one part of the reason for Googles success. But in such a situation, where Google tell their million of users that there are no paid hits mixed up with the organic results, and the users have to trust this when using the search engine, it would be morally wrong o
11、f Google to allow paid hits among the organic results anyway. There is no reason to believe that this actually happens because such a practice would go against Googles own business model. But my point is that it would also be a deceptive practice and hence morally problematic. The general algorithm
12、of some search engines, notably Googles PageRank, is of course well known. But Google uses in addition a large number of additional parameters in the indexing and ranking of web pages, which are not known in detail and which are probably changing. Hence, the question is whether all of the details ou
13、ght to be made public. Witten et al. sum up the dilemma nicely in their discussion of the search engines use of secrecy to fight spam. Users have a legitimate reason to know the recipes that govern the emergence of certain web pages from underneath a mountain of documents. What are you missing if yo
14、u inspect the first three, or the first 20 pages of the ranking? What if the ranking policy changes? The view of the web that thesearch engines present is arbitrary and changes mysteriously at unpredictable times. We are trapped in a dilemma: users want to know how their information is selected inde
15、ed, they have a right to but if this information were made public, spam would increase without bound, with disastrous consequences. And, of course, publishing the details of the ranking policy would create far more interest among spammers than among most ordinary users. Discussion On the analysis de
16、veloped here search engine companies are only morally required not to undermine the rational users search strategies. Others have argued that the moral duties of the search engine companies go farther than this. In a recent paper Introna and Nissenbaum have argued that the details of the algorithms
17、of the search engines should be made public. Their central argument for their suggestion is that search engines are public goods. The claim that the major search engines are public services is in my opinion clearly correct. They serve as central entry points for peoples access to information on the
18、web. It is also easy to agree with the suggestion that search engines should be thought of in political terms. Still, there are important objections to the suggestion that search engine algorithms should be made public. This is the serious risk that the publication of the details of the algorithms i
19、tself would create new bias. The problem is that web masters try to manipulate the search engines to improve their rank. And search engine spamming spamdexing is already a huge problem. Spam can take a number of forms. One classical method is link-farming, i.e. he creation of pages that link to your
20、 own target page to manipulate the search engines to give it a higher ranking. There is currently a war-like situation between the search engines and the so-called search engine optimizers. As the search engines continuously try to change their algorithms to prevent spamming, the spammers tries to f
21、ind out how to manipulate the service. In 2004 it was estimated that as much as 1015% of the web pages was spam. In 2003, it was even claimed by some bloggers that the central Google algorithm, PageRank, no longer worked because of spamming. It was claimed thatthe algorithm was no longer useful beca
22、use bloggers and CEOs had learned too much about it and had, in effect, changed the nature of the Web. Since PageRank is based on an optimistic assumption that all links are conceived in good faith with no ulterior motives, an assumption that no longer holds, then PageRank is no longer useful. Fortu
23、nately, PageRank is not dead, but the problem of spamdexing has caused a lot of effort to prevent manipulation. And the major search engines will in fact respond to serious attempts to manipulate the ranking by removing the site from its index. There is thus every reason to believe if the indexing a
24、lgorithms were made public, this problem would be so significant that the usefulness of the search engines would be considerably reduced. Nissenbaum and Introna mention this problem only to dismiss it: Search engine operators are loath to give out details of their ranking algorithms for fear that sp
25、ammers will use this knowledge to trick them. Yet, ethical Web-page designers can legitimately defend a need to know how to design for, or indicate relevance to, the ranking algorithm so that those who search find what is genuinely relevant to their searches. This response does not give the problem
26、the attention it deserves. The main problem with the response is that the manipulation of indexing is construed as a problem only for the search engine companies, i.e.that spammers will trick them. This is of course also true. But this is not the central problem if we see search engines as public se
27、rvices, as Introna and Nissenbaum of course do. Then the serious problem is that biased search engines provide less useful services to the public. This shows, I believe, that the search engine companies cannot be morally required to publish the details of their algorithms. However, I believe Introna
28、 and Nissenbaum are right in insisting on the value of open standards on the web. Their arguments lend support to initiatives like the Wikia Search-project which tries to develop an open source search engine on the model of Wikipedia. The initiator of this project is Jimbo Wales, the founder of Wiki
29、pedia. The motivation for this project is exactly to increase the transparency of search engines: Search is part of the fundamental infrastructure of the Internet. And, it is currently broken. Why is it broken? It is broken for the same reason that proprietary software is always broken: lack of free
30、dom, lack of community, lack of accountability, lack of transparency. Here, we will change all that. The idea is thus to develop a search engine that is based on human contributions to the ranking of search engine results, using an open algorithm. The project is still in its initial phases, and the
31、search engine performs very badly compared to the commercial search engines. It is an open question whether the project will succeed, but of course it would be a good thing if it did. But still this does not imply that all the other search engine companies are morally required to operate on the same
32、 model: the enforcement of such a requirement would mean that the quality of the existing search services would go down because if spam. This would be a loss not only for the searchers but also for small players on the web because on a web with more spam the findability of a web page would depend ev
33、en more than it does today on the resources the web site have available for search engine optimization. Hence, smaller and less resourceful web sites could end up being more marginalized. This would in my opinion also be a less democratic web. One suggestion might be that the algorithms of search en
34、gines could be controlled by a group of experts on behalf of the public. This would lift the veil of secrecy while voiding the problem of spam, it might be argued. Such an arrangement could perhaps make sure that the users are not deceived and that paid hits are not secretly allowed. But note that t
35、his would not meet the standards of openness advocated by Wales, and Introna and Nissenbaum because the details of the algorithm would still not be public knowledge. The arrangement would not provide information that could help users of the web to more rationally exploit its resources. Conclusions T
36、wo points emerge from this discussion. First, the big search engine companies are morally required to make their policies known to their users and to follow them. More generally, they are morally required to act in such a way that they do not undermine their users efforts to act as rational searcher
37、s. These moral requirements arise because of the important roles the big search engines play as providers of access to information on the web, and hence as contributors to the public use of reason. This conclusion is supported by considerations that are broadly Kantian in nature and seems to follow
38、if we apply his principles for the public use of reason to informational intermediaries like search engines. Second, on the analysis offered here the search engine companies are not required to publish the details of their algorithms. Two kinds of arguments are offered for this conclusion. One reaso
39、n is that the publication of the algorithm would make users worse of in terms of information. Hence, it is very likely that the flood of spam would effectively undermine the searchers rational strategies. However, it is not clear that the fact that the algorithms are not made public do undermine the
40、 users rational search strategies. 译文 搜索引擎和公众使用的原因 资料来源：伦理和信息技术， 2008（ 9） 作者： 埃里格斯 达格 搜索引擎已成为网络上访问信息的入口。在这个信息时代，谷歌和其他商业搜索引擎扮演着不可缺少的角色。我这里讲的重点是搜索引擎与用户的关系。基于搜索引擎排名并不会对用户公开，涉及到了一个权力不平衡的原则。大多数搜索引擎公司都是私人公司，并且提供免费服务，而每天都有成千上万的用户根据自己意愿，切换到不同的搜索引擎搜索信息，所以搜索引擎对用户不存在道德责任。但我认为这个观点是片面的。 关于搜索引擎 付费点击问题，谷歌和其他大多数搜索引
41、擎都会出现两个搜索结果 自然搜索结果和付费点击结果。例如，在谷歌搜索结果页面上，自然搜索结果一般列在页面的中部，而付费点击结果则列在页面的右侧或顶部。这种显示方式比较符合用户浏览习惯。但我认为谷歌或其他搜索引擎公司没有道德要求必须将结果分开。 20 世纪 90 年代后期， GoTo 公司创立了搜索引擎只基于完全支付的模式：在结果页面上的最好排名由竞价最高者所得。然而，事实上 GoTo 公司未能吸引很多用户，是因为搜索引擎付费点击并没有返回用户所期望的结果。不过，大多数用户只是了解搜索引 擎大致的排名情况，因此我觉得 GoTo 公司策略并没有错误。 虽然 GoTo 公司在 网络上销售广告存在一定
42、优势，但后来被谷歌收购，现在以竞价模式来销售广告，出现在谷歌搜索结果第一页的右侧。同时谷歌得出：用户不希望搜索引擎出现自然搜索和付费点击混合的结果。因此，谷歌选择了把这些不同类型的结果明确分开的战略。谷歌的分离式网页设计获得了成效。但在这种情况下，当谷歌告诉成千上万的用户有自然搜索和付费点击时，用户在使用时必须信任，而谷歌允许在自然搜索中混合付费点击，这将是道德上的错误。因为这种做法违背了谷歌本身的商业模式。但 我认为在某种程度上，它欺骗了用户，因此在道德上也有问题。 一些搜索引擎，特别是谷歌，它的排名算法是众所周知的。但谷歌用户除了在使用索引和网页的参数，细节是未知的，也是不能改变。因此，问
43、题的关键是是否所有的算法细节应该公开。 威滕 等人 对搜索引擎使用保密来对抗垃圾邮件的讨论有了很好的总结。 用户有理由知道管理一定网页的文件。如果你检查网页排名的前三页，或者前 二十项 看看是否缺少什么？如果排名政策改变会怎样？该网站认为，搜索引擎现在是武断的，在未知时间里是难以琢磨的。假如用户想知道如何选择自己的信息，事实上，他们也 有权利知道。但如果这些排名算法细节被公开，垃圾邮件必然将泛滥，造成灾难性后果。当然，相比于普通用户，公开排名算法细节这一政策将会给垃圾邮件制造者创造更多的利益。 讨论： 有人认为搜索引擎公司只是道德上的责任，不得破坏用户的理性搜索策略。还有人认为，搜索引擎公司的
44、道德责任远远不止这些。在最近的一篇文章中因切纳和 尼森鲍姆 围绕搜索引擎算法的细节是否应予以公开这一观点进行了一番争论。他们的中心论点是，搜索引擎是否是公共物品。 在我看来，那种宣称各大搜索引擎是公共服务，显然是正确的。搜索引擎是我们访问网络信息的入口点 。搜索引擎被认为是政治方面的建议，这一观点也很容易被接受。不过，搜索引擎算法应公开的建议引起极大的反对意见。该算法的细节公布本身也会造成新的偏见，这是严重的风险。然而，网站管理员试图操纵搜索引擎，提高他们的排名。搜索引擎垃圾邮件 垃圾索引已经是一个很大的难题。垃圾邮件可以采取多种形式。一种经典的方法是“交换链接”，比如某个网站的网页链接到自己
45、的目标网页，操纵搜索引擎给它创造一个较高的排名。目前存在搜索引擎和搜索引擎优化类似战争的局面。作为搜索引擎，不断尝试改变自己的算法，以防止垃圾邮件。而垃圾邮件制造者试 图找出如何操作搜索引擎的算法。据估计 2004 年，高达 1015%的网页是垃圾邮件。 2003 年，一些博客主甚至声称，因为垃圾邮件的泛滥，谷歌的中心算法，网页排名，不再有用，因为博客主和首席执行官已经学会了太多关于它的算法，并在实际上改变了网络的性质。由于网页排名是乐观假设各个链接都是友好的，且并没有恶意，但如果假设不成立，则网页排名就失效了。 幸运的是，网页排名并没有废弃，但垃圾索引问题已经引起了广泛关注，以防止操纵排名算
46、法。各大搜索引擎将积极响应，实际上通过索引来移动网站企图改变它的排名。因此，我们有充分 理由相信，如果索引算法公开，搜索引擎的作用将大大降低，这个问题将会如此引人注目。 尼森鲍姆和 因切纳提到这个问题，只能驳回。 搜索引擎运营商都不愿给出排名算法的细节，怕垃圾邮件制造者将利用这些理论来欺诈他们。然而，网页设计人员可以合法地保护自己需要知道的如何设计或说明相关的排名算法，使搜索者找到他们自己真正想要搜索的信息。 然而，这些反应的问题并没有给予应有的重视。反应的主要问题：只能是搜索引擎公司能操作索引，比如“垃圾邮件带有欺诈性”，这当然是正确的，因切纳和 尼森鲍姆 当然也这么认为，但作为公共服务的搜
47、索引擎， 这还不是中心问题。那么问题的关键是有偏见的搜索引擎为公众提供极少有用的服务。 搜索引擎公司没有道德责任要求公布其算法的细节。不过，我相信因切纳和 尼森鲍姆 是根据对网络的开放标准，坚持正确的价值。他们的论据支持了像维基亚的搜索项目，试图在维基百科的模式上开发一个开放式资源搜索引擎。该项目的发起人是神保威尔士，维基百科的创始人。这个项目的动机就是要增加搜索引擎的透明度。 搜索是互联网基础设施的一部分，但它当前被中断了。 为什么会中断呢？ 同样的原因，专有软件总是会中断：缺乏自由，缺乏团体，缺乏责任，缺乏透明度。在这 里，我们将改变这一切。 这个想法是建立这样一个搜索引擎 以开放式的算法
48、，基于对搜索引擎结果排名的人类贡献。该项目尚处于起步阶段，与商业搜索引擎相比，这样的搜索引擎表现得非常糟糕。该项目是否会成功是尚未解决，如果该项目成功了，当然必将是一件好事。但这样做并不意味着所有搜索引擎公司有道德责任需要运行相同的模式：但因为垃圾邮件的存在，这项规定的执行意味着现有的搜索服务质量会下降。 这不仅是对搜索者，对网络上的用户来说都将是一个损失。原因是对搜索引擎优化来说，网络上有更多可发现的垃圾邮件将更依赖于网络可用的资源。因此，规模小，财力稍逊的网站最终可能会更加边缘化。在我看来，这也将是一个不太民主的网站。 在此我提出了一个可行的建议：搜索引擎的算法可以通过由代表公众的专家小组
49、控制。而垃圾邮件的问题，也许会引起争辩。这种安排也许可以确保用户不会受骗，而付费点击也不会暗里地被允许。但是请注意，这将不符合威尔士倡导开放的标准，因切纳和 尼森鲍姆 因为算法的细节仍不会公诸于众。这样 他们的用户和跟随他们的政策。通常，他们确实必须以这样的方式行事，他们的安排对用户来说不会提供更合理资源的信息。 结论： 从这一讨论中显露出两点：首先，大 的搜索引擎公司确实需要让人们了解不会破坏用户的努力来担当理性的搜索者。这些道德责任的出现是因为大的搜索引擎公司在网络上扮演着信息提供者的角色，因此它们是作为贡献者为公众使用。像搜索引擎一样，作为信息中介，通过康德性质加上遵循运用他的理论为公众使用。 其次，以上分析了搜索引擎公司并没有要求公布其算法的细节。结论出现两种不同的争论。其中一个争论是：该算法的公开，会使用户获取的信息更糟糕。因为很有可能垃圾邮件制造者知道了这种算法，导致垃圾邮件泛滥，破坏搜索合理有效性的战略。不过算法不公开是否会破坏用户的理性搜索策 略，目前尚不清楚。
