True Causes
ZH
Bahasa IndonesiaDanskDeutschEnglishEspañolFrançaisItalianoNederlandsNorskPolskiPortuguêsSrpski (latinica)SvenskaTiếng ViệtTürkçeΕλληνικάРусскийСрпски (ћирилица)Українська中文日本語한국어
← 大厅
7/7 · 行动
30 人 · 32 个想法
点击任意已完成的阶段进行查看
✓
提出
✓
澄清
✓
归类
✓
选择
✓
结构
✓
地图
7
行动
演示 · 不计入
障碍 · 持有链接的任何人 · 公开列出 · 已开启联盟
主持人：它的主持人（已封存）
30 人 · 32 个想法 · 2 个小组 · 3 个房间 已结成联盟
通行证 → · 逻辑 → · 记录
↗ 分享这张地图
在其他网站上，作为实时地图：
作为图片： https://truecauses.org/d/55824a6/map.svg
作为数据： 本房间每一个已判断的卡片对（CSV）. 每个主张都有自己的地址 /c/55824a6/<card>，列在地图下方。
发布时请注明出处： Causal map: True Causes, https://truecauses.org/d/55824a6
问题

Why do we stay inside echo chambers?

提出该做什么，再判断每项行动能否撼动每个根源。

封存于 19 天前：这个房间已不再知道谁曾在其中；只有主持人仍然留名。它得出的结论保留。
30 人在座 （其中 29 名为模拟成员） · 针对 4 个根本原因已提出 3 项行动。添加一项，然后判断每项行动能否撼动每个原因。
正以访客身份查看。输入用户名即可加入并参与。
有些主张在一个循环中相互影响
↺ We built for a scale nobody had governed befor → Nobody agrees on who should decide what is tru → Ad money rewards volume of attention, not qual
循环可能是一项真实的发现——一个自我强化的回路——也可能是一条方向从未真正确定的关联。地图把这组主张视为一个节点，所以其中没有一项单独是根源。如果是后一种情况，值得把这个卡片对重新交给房间。
先给根本原因打分
为什么
本方法自身的做法，在提出任何行动之前：每个根本原因按 1–5 分打分——解决它的可行性有多高、解决它会带来多大影响、不加干预它自行解决的可能性有多大。优先考虑影响大、可行性高、自行解决可能性低的根源——优先级 = 影响 + 可行性 − 可能性。分数在小组内取平均，并显示有多少人打了分。
1 = 低，5 = 高。这是你自己的分数；旁边是小组的平均分。
根本原因可行性影响自行解决优先级
The metric was chosen because it was easy to log––––
Ranking optimises for time spent, not for being informed––––
Regulation asks for transparency reports nobody can verify––––
Researchers ask for data and legal says no––––
每项行动能撼动每个根本原因吗？
竖列：大家提出的行动。横行：地图找出的根本原因。每个格子都是对一个问题的一次投票——这项行动能否显著减少那个原因？一方必须达到所投票数的 75% 才算定下；达不到，格子就保持分裂。 每个格子显示其投票分布——是 / 更糟 / 否——当其中一项达到四分之三时，该格子就定下了。
为什么
以所投票数的 75% 定下，和这里的其他所有问题一样。分裂表示有人投了票，但没有一方达到这个标准——这是一个结果，不是排队等待，它会一直保持分裂，直到足够多的人改变主意。没有人定下的格子保持未定；它永远不会变成否。逐对单独判断，能防止一项受欢迎的行动被记上它根本没触及的功劳。
行动根本原因 · 左边的行动能减少它吗？
The metric was chosen because it was easy to logRanking optimises for time spent, not for being informedRegulation asks for transparency reports nobody can verifyResearchers ask for data and legal says no
Give safety work a metric and headcount 讨论✗ 不，不能
4 是 / 0 更糟 / 25 否
尚未确定
⚑ 小组意见分裂
8 是 / 0 更糟 / 21 否
尚未确定
Independent audit access under a legal safe harbour 讨论✓ 是，它能减少这一项
27 是 / 0 更糟 / 2 否
尚未确定
✗ 不，不能
5 是 / 0 更糟 / 24 否
尚未确定
Publish a ranking change log anyone can read 讨论✗ 不，不能
6 是 / 0 更糟 / 23 否
尚未确定
✗ 不，不能
2 是 / 0 更糟 / 27 否
尚未确定
目前杠杆作用最大
Independent audit access under a legal safe harbour 减少了 4 个根本原因中的 1 个
地图上的行动
显示的是房间判断过的所有行动：它直接作用于 4 个根本原因中的 1 个。 有 3 个没有直接作用。
只依据房间在行动阶段定下的格子，别无其他。绿色的根本原因至少被所显示行动中的一项减轻；灰色的，没有一项直接作用于它；红色边框表示其中某项会使它更糟。
分享此视图： 作为图片 · 作为供其他网站使用的实时地图
Regulation asks for transparency reports nobody can verify → People choose the sources that already agree with themThere is no reviewer who speaks the language of half the reports → Automated removal is cheaper than being rightRanking optimises for time spent, not for being informed → Researchers ask for data and legal says noRanking optimises for time spent, not for being informed → A moderator sees the worst of it for the wage of a call centreWe built for a scale nobody had governed before → Nobody agrees on who should decide what is trueEvery fix ships to the users who complain loudest → We cannot publish what we remove without teaching people to evade itEvery fix ships to the users who complain loudest → The people affected worst are in markets we do not staffEvery fix ships to the users who complain loudest → Ad money rewards volume of attention, not quality of itThe metric was chosen because it was easy to log → Ranking optimises for time spent, not for being informedThe rules are written in one country and applied in forty languages → There is no reviewer who speaks the language of half the reportsResearchers ask for data and legal says no → Recommendation systems are not inspectable from outsideResearchers ask for data and legal says no → The rules are written in one country and applied in forty languagesThe rules are forty pages and the decision takes twelve seconds → Recommendation systems are not inspectable from outsideThe rules are forty pages and the decision takes twelve seconds → Automated removal is cheaper than being rightPeople choose the sources that already agree with them → Every fix ships to the users who complain loudestA moderator sees the worst of it for the wage of a call centre → The rules are forty pages and the decision takes twelve secondsNobody agrees on who should decide what is true → Ad money rewards volume of attention, not quality of itNobody outside can reproduce a single claim we make → Nobody agrees on who should decide what is trueAd money rewards volume of attention, not quality of it → We built for a scale nobody had governed beforeRegulation asks for transparency reports nobody can verifyRegulation asks fortransparency reports nobodycan verify所显示的行动对这个根本原因的作用没有行动直接作用于它The metric was chosen because it was easy to log ② Independent audit access under a legal safe harbourThe metric was chosenbecause it was easy to log②所显示的行动对这个根本原因的作用1 项行动减轻它Nobody outside can reproduce a single claim we makeNobody outside canreproduce a single claim wemakeRanking optimises for time spent, not for being informedRanking optimises for timespent, not for beinginformed所显示的行动对这个根本原因的作用没有行动直接作用于它People choose the sources that already agree with themPeople choose the sourcesthat already agree withthemA moderator sees the worst of it for the wage of a call centreA moderator sees the worstof it for the wage of acall centreResearchers ask for data and legal says noResearchers ask for dataand legal says no所显示的行动对这个根本原因的作用没有行动直接作用于它Every fix ships to the users who complain loudestEvery fix ships to theusers who complain loudestThe rules are forty pages and the decision takes twelve secondsThe rules are forty pagesand the decision takestwelve secondsThe rules are written in one country and applied in forty languagesThe rules are written inone country and applied inforty languagesThe people affected worst are in markets we do not staffThe people affected worstare in markets we do notstaffWe cannot publish what we remove without teaching people to evade itWe cannot publish what weremove without teachingpeople to evade itAd money rewards volume of attention, not quality of itAd money rewards volume ofattention, not quality ofit属于一个反馈回路⟳Nobody agrees on who should decide what is trueNobody agrees on who shoulddecide what is true属于一个反馈回路⟳We built for a scale nobody had governed beforeWe built for a scale nobodyhad governed before属于一个反馈回路⟳Recommendation systems are not inspectable from outsideRecommendation systems arenot inspectable fromoutsideThere is no reviewer who speaks the language of half the reportsThere is no reviewer whospeaks the language of halfthe reportsAutomated removal is cheaper than being rightAutomated removal ischeaper than being right
地图上的每个编号标出该行动所减轻的一个根本原因。
① Give safety work a metric and headcount
② Independent audit access under a legal safe harbour
③ Publish a ranking change log anyone can read
我们预期接下来会发生什么
关于各项行动的预测：每项应改变什么、如何衡量、何时检查。到了那一天，曾在房间里的某个人会在旁边写下实际发生了什么。
为什么
地图说明哪些行动能撼动哪些原因。它不说明是否有人做了这些行动，也不说明是否有什么改变了。这里的每一条都是一项预测，带有日期和要衡量的东西，在地图定下后写成。日期到了，实际发生的情况会写在旁边，而预测保持原样：错误的预测才是有用的那种。没有人对这些投过票，它也不改变房间发现的任何东西。
到期：24 Mar 2027
行动Publish a ranking change log anyone can read
预期a change to what people see is visible to the people who see it
衡量指标% of ranking changes appearing in a public log within a month
估计房间：40（15 人的中位数，范围 15 到 70）