Data, algoritma, dan kekuatan digital
International Baccalaureate · IB Diploma · Digital Society · HL Topik 1 3:56 Narasi bahasa Inggris · Subtitle bahasa Inggris + 中文 disematkan langsung
Bab-bab
Transkrip
In 2020 the pandemic cancelled England's A-level exams, so a formula decided the grades.
2020 年疫情取消了英格兰的 A-level 考试，于是由一个公式来决定成绩。
It took each school's past results and each teacher's ranking of their students, and fitted this year's class to the old pattern.
它用每所学校过去的成绩和每位老师给学生的排名，把今年的学生套进旧的模式。
Almost forty per cent of the grades teachers had predicted were lowered.
老师预估的成绩中，近百分之四十被调低。
Students protested against the algorithm, and four days later the government dropped it.
学生上街抗议这个算法，四天后政府放弃了它。
That is what digital society studies.
这正是数字社会要研究的。
The course has three columns of seven.
课程有三列，每列七项。
Concepts are the questions you ask, content is the technology you study, and contexts are the parts of life where it matters.
概念是你提出的问题，内容是你研究的技术，情境是它影响的生活领域。
You never study the twenty-one one by one; every inquiry takes one or two from each column and joins them to a real case.
你不会把这二十一项逐个学；每次探究都从每列取一两项，再连到一个真实案例上。
A concept is not a topic to describe; it is a question you ask of a case.
概念不是要描述的话题，而是向案例提出的问题。
Ask power of the 2020 grades and you get who decided, teachers, the exam board or the formula's designers.
用"权力"去问 2020 年的成绩，就得到：是谁决定的，老师、考试局，还是公式的设计者？
Ask values and ethics and you get whether it is fair to judge a person by other people's past results.
用"价值观与伦理"去问，就得到：用别人过去的成绩来评判一个人，公平吗？
Data are raw facts.
数据是原始事实。
A smartwatch records seventy-two, seventy-five, one hundred and forty.
智能手表记录下 72、75、140。
Information is that data organised to mean something, your heart rate rose sharply during your run.
信息是把这些数据组织起来、让它有意义：你跑步时心率急剧上升。
And every dataset has makers, who decided what to collect, how to process it, and who stores it.
每个数据集都有制作者，由他们决定收集什么、如何处理、由谁保存。
In 2014 about two hundred and seventy thousand people took a personality quiz on Facebook.
2014 年约有二十七万人在 Facebook 上做了一个性格测试。
But it also collected their friends' data, so up to eighty-seven million people's data reached Cambridge Analytica, which used it to target political advertising.
但它还收集了他们好友的数据，结果多达八千七百万人的数据落到剑桥分析公司手里，被用来定向投放政治广告。
Most of them never used the app.
其中大多数人从没用过这个应用。
Data collected for one purpose was used for another.
为一个目的收集的数据被用在了另一个目的上。
An algorithm is a set of step-by-step rules that turns inputs into outputs.
算法是把输入变成输出的一组逐步规则。
The key idea is bias in, bias out.
关键是"偏见进，偏见出"。
The grading formula used each school's past results, so students at weaker schools were held down.
成绩公式用了各校过去的成绩，所以基础较弱学校的学生被压低。
Amazon trained a hiring system on ten years of CVs, mostly from men, and it learned to mark down the word women's.
亚马逊用十年的简历训练招聘系统，这些简历大多来自男性，结果它学会了给含有"女子"一词的简历打低分。
A focused inquiry joins all four parts. Power is the concept, algorithms the content, social and political the contexts, and the 2020 grading formula the real-world example.
聚焦的探究要把四部分连在一起：概念是权力，内容是算法，情境是社会与政治，真实例证是 2020 年的成绩公式。
Together they ask one question. Did the formula move the power to decide a grade from teachers to the formula's designers?
合起来只问一个问题：这个公式是否把决定成绩的权力，从老师手里转到了公式设计者手里？
Drop one part and the inquiry fails.
少了任何一部分，探究就会失败。
Concept only is an essay about power with no real case.
只有概念，就是一篇空谈权力、没有真实案例的文章。
Case only is the story of 2020 with no concept to explain it.
只有案例，就是讲 2020 年的故事，却没有概念去解释。
And a context dump says a lot about schools and politics but never shows a digital system.
情境堆砌则大谈学校和政治，却看不到任何数字系统。
Paper 1 runs from short answers to extended responses.
试卷一既有简答也有长答。
Identify and outline want short, exact answers.
identify 和 outline 要求简短准确。
Examine, discuss and evaluate want a long answer that defines the terms, names the concept, content and context, explains a real case in detail, gives more than one perspective, and ends with a judgement.
examine、discuss 和 evaluate 要求长答：先定义术语，点明所用的概念、内容和情境，详细解释一个真实案例，给出不止一种观点，最后下结论。
Here is a plan for a describe question about young people's privacy.
这是一道关于青少年隐私的 describe 题的答题计划。
The context is political.
情境是政治。
The example is Cambridge Analytica, where a quiz collected data about friends who never agreed.
例子是剑桥分析事件，一个测验收集了从未同意的好友的数据。
The concern is that young people share likes and friend lists freely.
担忧在于青少年随意分享点赞和好友列表。
And the link is data moved to a new purpose, a question of power and of ethics.
最后的联系是：数据被挪作他用，这是权力和伦理的问题。
Check yourself without notes. Ask power and ethics of the 2020 formula, separate data from information, explain bias in, bias out with two cases, and build one focused question of your own.
不看笔记自测：用权力和伦理去追问 2020 年的成绩公式；用一个例子区分数据和信息；用两个案例解释"偏见进，偏见出"；再自己构建一个聚焦的问题。