My company book club is currently reading an important one: Algorithms of Oppression. This has sparked strong interest — in myself as well as many coworkers — to take action. I encourage you to read the book as well, to learn more about the core problems inherent to private algorithms. Ideally, algorithmically-driven decisions should represent all people equitably… but that’s currently not often the case.

我公司的读书俱乐部目前正在阅读一个重要的书： 压制算法 。 这引起了我本人以及许多同事的强烈兴趣，采取行动。 我鼓励您也阅读本书，以了解有关私有算法固有的核心问题的更多信息。 理想情况下，算法驱动的决策应该公平地代表所有人……但是目前并不常见。

搜索引擎如何增强种族主义。 (How Search Engines Reinforce Racism.)

Just as our world evolves, this blog will function as my personal evolving account of a shared journey towards reducing algorithmic oppression. It’s a small first step, at the moment: recognizing the problem, and that there are easy ways for every human who can read this message to help, right now.

就像我们的世界在发展一样，此博客将作为我个人在不断发展的减少算法压迫的旅程中不断发展的描述。 目前，这只是很小的第一步：认识到问题，并且对于每个阅读此消息的人来说，都有简便的方法可以为您提供帮助。

This first post lays out initial plans for counteracting algorithmic oppression — a pervasive problem in current technology, especially prevalent in search engines. Even with equality at their core, evolving algorithms quickly become complex, and have the potential to oppress certain groups or individuals unfairly. This can be prevented.

第一篇文章列出了抵消算法压迫的初步计划，这是当前技术中普遍存在的问题，尤其是在搜索引擎中普遍存在。 即使以平等为核心，不断发展的算法也会Swift变得复杂，并有可能不公平地压迫某些群体或个人。 这是可以避免的。

Perspective — for example, seeing digital trust and personal trust as separate yet related issues — is important to keep. Open source, repeatable digital trust algorithms are just one way to help fix inequity, without physical contact or risk involved. These new algorithms and patterns have the powerful potential to be new standards for personal digital content verification and filtering.

保持远见非常重要，例如，将数字信任和个人信任视为独立但相关的问题。 开源，可重复的数字信任算法只是解决不平等问题的一种方法，而无需涉及物理联系或风险。 这些新的算法和模式具有强大的潜力，成为个人数字内容验证和过滤的新标准。

算法均衡器当前有两个任务： (Algorithm EQ has two current missions:)

1. Publish information about existing inequitable algorithms, and potential methods for improving them, via communication through many existing platforms.

   通过许多现有平台上的通信，发布有关现有不平等算法的信息以及改进这些算法的潜在方法 。

2. Create simple, intuitive apps enabling personalized content searching and filtering in equitable ways— without relying on private external algorithms. These apps are still in a modest, “digital back-of-the-napkin” status, and we are open to all constructive feedback.

   创建简单，直观的应用程序，以公平的方式实现个性化内容搜索和过滤，而无需依赖外部私有算法。 这些应用仍处于适度的“数字化餐巾纸”状态，我们对所有建设性反馈持开放态度。

更平衡的方法 (A more balanced approach)

Instead of relying on privatized content sorting and filtering, or privately-funded media promotion (e.g. AdWords), these open source principles can be applied to surface more useful and equitable content to users. To be clear, implementing and using these open source methods and apps are a choice. Users have the right to ignore them, and may continue using existing private algorithmically-filtered and sorted portals instead (read: Google). Those private algorithms are not going anywhere, and that’s ok! Many companies have already started improving their existing algorithms, based on the type of feedback our group and others have been giving, and that’s also a good thing. It’s just not enough, on its own.

这些开放源代码原则可以不必依赖私有化的内容分类和过滤或私人资助的媒体推广(例如AdWords)，而可以向用户展示更有用和公平的内容。 需要明确的是，实现和使用这些开源方法和应用是一种选择。 用户有权忽略它们，并可以继续使用现有的经过算法过滤和排序的私有门户(请参阅：Google)。 这些私有算法什么都没有用，没关系！ 许多公司已经根据我们团队和其他公司提供的反馈类型开始改进其现有算法，这也是一件好事。 仅靠它本身还不够。

算法的公平性-对您，我和Olivæ (Algorithm equity — for you, me, and Olivæ)

Our platform has the potential to reduce algorithmic oppression significantly, by reducing reliance on private external algorithmic decisions, namely: filtering, sorting, and promoting specific digital content and signals. Based on shared, encrypted binary decisions between you and all your trusted groups of humans, with their own personal trust keys and chains, we can all grow together into an equitable algorithmic future of trusted blockchain-style content verification signals.

通过减少对私有外部算法决策的依赖，即过滤，排序和推广特定的数字内容和信号，我们的平台具有显着降低算法压迫的潜力。 基于您和您所有可信赖的人类群体之间的共享，加密的二进制决策，以及他们自己的个人信任密钥和链，我们可以共同成长为可信赖的区块链式内容验证信号的公平算法未来。

奥利维？ (Olivæ?)

与切线相关的数字发现 (A personal aside, of tangentially-related digital discovery)

This is a new name, created for an important future human— my first child. My fiancé Saxy and I are planning on giving this alias (i.e. middle name) to our child, due in mid November 2020. Interestingly, the character æ is not stored or displayed correctly by many legacy digital systems. This brings us to our first side quest — feel free to skip this section if already a character-encoding pro, or simply not interested.

这是一个新名字，是为一个重要的未来人类-我的第一个孩子而创建的。 我和我的未婚妻萨克斯(Saxy)计划在2020年11月中旬给孩子使用这个别名(即中间名)。有趣的是，许多传统数字系统并未正确存储或显示字符æ。 这使我们开始了第一个任务-如果已经是字符编码专家，或者根本不感兴趣，请随时跳过本节。

æ (æ)

This character, the ending of Olivæ’s new alias, is easily typed on a modern mac: while holding the option key, press ’ (apostrophe, next to return) to type it: æ. See? With a Windows system it’s a bit trickier, but usually not too hard to type either: alt-145 for example (but there are caveats). See this Wikipedia article and its sources for extra nerdy details, if you’re curious about non-ascii character encoding quirks.

这个字符，是Olivæ新别名的结尾，很容易在现代Mac上键入：按住Option键的同时，按'(撇号，返回旁边)以键入它：æ。 看到？ 对于Windows系统，这有点棘手，但通常都不难键入：例如alt-145(但有一些警告)。 如果您对非ASCII字符编码的怪癖感到好奇，请参阅此Wikipedia文章及其来源以获取更多书呆子详细信息。

Æ (Æ)

The capitalized version of this same character may look familiar, as it was recently used within a certain new celebrity child’s even more unique name: X Æ A-12 Musk. Joke, or not-joke? I am personally still not sure… and would love to hear a definitive answer on that directly from the sources. As far as we all can tell, its proper display and pronunciation is still up for debate!

这个人物的大写字母可能看起来很熟悉，因为它最近在某个新的名人孩子的甚至更独特的名字中使用： XÆA-12 Musk 。 开玩笑还是不开玩笑？ 我个人仍然不确定……并且很想直接从消息来源中听到一个明确的答案。 据我们所知，其正确的显示和发音仍有待争论！

投票链 (Votechain)

Alright, back to the main topic at hand: algorithmic equity. I’d love to gather and tally everyone’s opinions about this æ character, and whether it’s a good idea to use in a human’s name. It looks pretty cool to me, and we’re both excited to share it with our future child. But are we completely off-base? Are there reasons we haven’t thought of, why it might not be a great idea? We’ve still got a few months to decide before it’s official. This is where you come in!

好了，回到当前的主要主题： 算法公平 。 我很乐意收集并收集每个人对这个角色的看法，以及以人名命名是否是一个好主意。 对我来说，它看起来很酷，我们都非常高兴与我们未来的孩子分享它。 但是我们完全脱离基地了吗？ 我们有没有想到的原因，为什么它可能不是一个好主意？ 在正式发布之前，我们还有几个月的时间来决定。 这就是你进来的地方！

Public votechains like this can be easily created and posed as questions to vote on, and even optionally include near real-time text, video, and/or audio discussions, between known and verified chain participants. Unlike a traditional paper vote, which is a singular object you cast once per term, a votechain is a living, digitally-breathing aggregated result — the community’s pulse on the issue. You can change your own easily (daily, for now) up until the final tally, based on any new discussion or information that sways you.

这样的公共投票链可以很容易地创建并提出要投票的问题，甚至可以选择在已知和已验证的链参与者之间进行近乎实时的文本，视频和/或音频讨论。 与传统的纸质投票不同，传统的纸质投票是您每学期只投一次票，而投票链则是一种生动的，数字化的汇总结果，是社区对此事的关注。 您可以根据影响您的任何新讨论或信息，轻松地(每天一次)更改自己的内容，直到最终计数。

Personal votechain values are submitted and updated privately, encrypted and validated with at least two public-private keypairs, used both to encrypt your personal information and confirm validity between the existing votechain and your new individual choice.

个人投票链的值是私下提交和更新的，并使用至少两个公共-私人密钥对进行加密和验证，用于加密您的个人信息并确认现有投票链和新的个人选择之间的有效性。

For any votechain-based decision, validated individual votes are quickly and accurately tallied, aggregated based on each individual’s unique vote ID or alias. Only each user’s most recent is counted, if multiples submitted — based on a simple timestamp. Every participant has the capability to verify the accuracy of their vote and the votes of their trusted peers in these tallies, as the ID or alias of every vote is public (but not any other identifying details).

对于任何基于投票链的决策，都会根据每个人的唯一投票ID或别名快速准确地对经过验证的个人投票进行汇总。 如果提交了多个副本，则仅计算每个用户的最新身份(基于简单的时间戳记)。 由于每个投票的ID或别名都是公开的(但没有其他任何标识性详细信息)，因此每个参与者都有能力验证其投票和受信任同级投票的准确性。

我们如何做得更好 (How We Can Do Better)

确定我们(包括您在内)现在可以做的基本工作，以帮助提高算法公平性，同时确定并实施其余的信任链和投票链实施细节。 认真聆听所有初始反馈，以确保第一个公开测试版有用且公平。 (Identify basic things we (including you!) can do now to help increase algorithmic equity, while deciding on and implementing the remaining trust-chain and vote-chain implementation details. Ensure the first public beta is useful and equitable, by listening to all initial feedback carefully.)

My employer, UrbanSitter, has taken action already in this first regard, by improving their internal user moderation patterns and search software to be more equitable. I personally trust and admire everyone involved with this company: a strong and caring female executive staff, and board, that is refreshingly and admirably equitable in many ways already — and still striving to do even better.

我的雇主UrbanSitter已经在第一方面采取了行动，通过改善其内部用户审核模式和搜索软件使其更加公平。 我个人信任并钦佩与该公司有关的每个人：一个强大而有爱心的女性行政人员和董事会，在许多方面已经令人耳目一新，而且令人钦佩，而且仍在努力做得更好。

It’s easy to feel overwhelmed, when realizing how much work we have ahead. We can’t fix every existing algorithm today, or even this year. Our software improvements and betas briefly described earlier barely scratch the surface, revealing deeper and more complex algorithmic bias below. But, we can start small — starting now — by all doing better where we can.

当意识到我们要完成多少工作时，很容易感到不知所措。 今天甚至今年我们都无法修复所有现有算法。 前面简要介绍的我们的软件改进和Beta版几乎没有触及表面，下面揭示了更深，更复杂的算法偏差。 但是，我们可以从小处着手-从现在开始-尽一切可能做得更好。

我们的下一步 (Our Next Steps)

My own next step, in process now by publishing this post: find and assist a willing group of beta participants outside our existing small social circles of alpha participants. Continue practicing what we preach, by identifying and improving existing algorithms whenever possible to increase human equity.

我自己的下一步，现在正在通过发布此帖子来进行：在我们现有的小型alpha参与者社交圈之外，找到并协助一群自愿的beta参与者。 尽可能地通过识别和改进现有算法来提高人的公平性，继续实践我们的讲道。

Most of us plan to continue working at existing companies which run for profit, while not receiving any salary or other financial incentives for developing our new open source technology or increasing the equity of existing algorithms. None of our private companies currently — but may eventually (soon, I hope!) — use the open source algorithms or applications developed and published by our organization, Algorithm EQ.

我们大多数人都计划继续在以盈利为目的的现有公司工作，而不会因开发我们的新开源技术或增加现有算法的权益而获得任何薪水或其他财务激励。 目前，我们的私人公司都没有一家，但最终(希望很快)会使用我们的算法EQ组织开发和发布的开源算法或应用程序。

下一步 (Your Next Steps)

We’ve created helpful accounts to follow for tips and discussions: LinkTree, Instagram, Twitter, and Facebook. We also created a basic Algorithm EQ website, where we plan to share actionable guides to help you improve algorithmic equity, and eventually beta versions of our software.

我们创建了有用的帐户来跟踪提示和讨论： LinkTree ， Instagram ， Twitter和Facebook 。 我们还创建了一个基本的Algorithm EQ网站 ，我们计划在其中共享可操作的指南，以帮助您提高算法的公平性，并最终改善我们软件的beta版本。

Side quest: we repurposed a friend’s fun old Twitter account for this project. Its purpose was originally to explain and discuss menus for unique local lunch spots, e.g. the Kitchenette SF loading dock (my personal favorite at the time). You’re welcome to like/share/reply, retweet those, or simply read more recent haikus.

附带要求：我们将朋友的有趣的旧Twitter帐户用于该项目。 其目的最初是为了解释和讨论独特的当地午餐点的菜单，例如Kitchenette SF装卸平台(当时我个人最喜欢)。 欢迎您喜欢/分享/回复，转发这些推文，或只是最新的句。

我们都刚刚开始 (We’re all just getting started)

There have already been great people working on reducing inequitable algorithmic bias for some time now (thank you!) — yet there’s so much more to do. We’re reaching a cultural tipping point, and it’s finally something that our society is ready to change. How can you start helping? For now, let’s slow down and drop into the coda of this first post, by ending with haiku; a thing I personally like to do:

现在已经有很多人致力于减少不公平的算法偏差(谢谢！)—但是还有很多事情要做。 我们正在达到文化的临界点，这最终是我们的社会准备改变的事情。 您如何开始提供帮助？ 现在，让我们放慢脚步，以haiku结尾结束这篇文章的尾声。 我个人喜欢做的事情：

nothing digitalis ever truly staticalways evolving

从来没有真正的数字化总是永远发展