概述
题图来自互联网
由于缺乏信息源或某种主观意识,我们很容易陷入信息偏差,进而影响当前的决策。然而,我们可能能够感受到信息偏差的存在,但我们可能不知道这种信息偏差是如何产生的。
[En]
Due to the lack of information sources or some kind of subjective consciousness, we are easy to fall into information deviation, and then affect the current decision-making. However, we may be able to feel the existence of information bias, but we may not know how this information bias comes into being.
在《当下的启蒙》一书中,作者提到了一种常见的信息偏差。在两个人均经济发展水平相差无几的地区,A 地区比 B 地区多了一倍多人口,两个地区的恶性事件发生比例差不多,但是 A 地区的人们却会有种生存不下去的感觉。由于 A 地由于人口比 B 地区多,媒体会追逐那些吸引眼球的新闻,使得从媒体上看,恶性事件的实际数量比 B 地区多一倍。这就是由于传播造成的一类典型的信息偏差。
人与人之间最大的不平等,就是信息不对称。作为互联网时代发展的受益者,看似紧跟时代潮流的 IT 技术圈,从业人员几乎都是高层次的人才,或许应该会有着比传统行业更完善的信息评价体系,似乎应该更能从海量信息中提取出符合自己需求的信息,但实际上由于各种因素的存在,信息偏差发生得更加显著。
总结起来,这些影响因素包括"熵"、媒体关注度及"大厂回音壁效益"、技术圈子追逐热点,大数据推荐算法带来的"过滤气泡"等,多重因素交集,形成了当下 IT 圈信息传播的特点。
熵
"熵"起初来源于热力学,1948 年 C.E.Shannon(香农)热力学中将此概念借用过来,用以解决对信息的量化度量问题,并称为"信息熵"。信息熵是信息论中用于度量信息量的一个概念,总结而言:一个系统越是有序,信息熵就越低;反之,一个系统越是混乱,信息熵就越高。在传播中是指信息的不确定性,一则高信息度的信息熵是很低的,低信息度的熵则高。同样,信息熵也符合热力学第二定律:热量总是从热量高的地方向热量低的地方传输。
用物理学家薛定谔来说,"自然万物都趋向从有序到无序,即熵值增加(熵增)。而生命需要通过不断抵消其生活中产生的正熵,使自己维持在一个稳定而低的熵水平上(熵减)。生命以负熵为生。"
熵也可以用来描述一些未知的知识领域。如果将这些区域细分,可以分为舒适区、学习区和恐慌区。由于每个人都有自己的认知层,他们总是优先看到自己舒适区内的信息,习惯于用自己的舒适区认知来理解非舒适区的信息。
[En]
Entropy can also be used to describe some unknown areas of knowledge. if these areas are subdivided, they can be divided into comfort zone, learning area and panic area. As everyone has their own cognitive layer, they always give priority to seeing the information within their comfort zone, and are accustomed to using their own comfort zone cognition to understand the information of non-comfort zone.
其实,要理解高熵的知识,成本是非常高的,如果我们不经常梳理我们的认知,那么我们的认知就不会得到提高,最终只能被腐化,陷入无序的最大熵。当一个人的认知模型的熵增已经很大时,它自然会受到其认知模型的误导,从而影响信息的提取,形成信息偏差。
[En]
In fact, to understand high-entropy knowledge, the cost is very high, if we do not often sort out our cognition, then our cognition will not be improved, and finally can only be corrupted, into the maximum entropy of disorder. When the entropy increase of a person's cognitive model has been very large, it will naturally be misled by its cognitive model, thus affecting the extraction of information and forming information deviation.
媒体和回音壁效益
媒体总是只关心那些吸引眼球的热点新闻,甚至夸大一些新闻。在《事实》一书中,作者介绍了一起北极熊杀死猎人的案例,并获得了大量媒体报道,而在另一起家庭暴力事件中,一名妇女被前夫用斧头杀害,但从未得到媒体报道。在瑞典,熊是百年一遇的罕见事件,而家庭暴力导致的死亡几乎每隔一段时间就会发生一次,差距为1300倍。因此,熊是大新闻,但家庭暴力不是。
[En]
The media always only care about those hot news that attract attention, and even exaggerate some news. In the book Facts, the author introduces a case in which a polar bear kills a hunter and gets a lot of media coverage, while in another domestic violence, a woman is killed by her ex-husband with an axe but never gets media reports. In Sweden, bears are rarely seen in a century, while deaths caused by domestic violence occur almost every once in a while, a gap of 1300 times. So the bear is big news, but domestic violence is not.
科技类媒体似乎也难以脱俗,如最近 Facebook 改名为 Meta,并带来了元宇宙和 NFT 的繁荣,以及马斯克在太空探索、新能源领域取得的某些成就,总是一遍又一遍出现在我们的视野范围内,事实上这些东西离我们还非常遥远,除了满足猎奇心,短期内不会产生效应。
而把时间再拨回到 2018-2019 年左右,那时占据头条的信息几乎都是互联网新金融创新或区块链创新,彼时的开发者们几乎人手一本"区块链开发教程",仿佛不学区块链就是老年程序员一般。当然,这些热潮过后,留下了一地鸡毛,也是一代人的共同记忆。
除了某些新概念带来的热点,还有与大厂有关的常规新闻,如美国的 FAAG,中国的前 BAT。这些互联网大厂经常会时不时搞出一些技术资讯,如某些概念技术的运用,从而引发一些热点。随着各类科技媒体竞相宣传,并创造出了一个个"回音壁效益"。——回音壁效应这种概念,是指"在一个相对封闭的环境上,一些意见相近的声音不断重复,并已夸张或其他扭曲形式重复,令处于相对封闭环境中的大多数人认为这些扭曲的故事就是事实的全部"。
在现代社会,由于互联网的应用和社交媒体的发展,这种现象更加深刻,因为一些商业网站根据搜索结果的记录提供性质相似的网站材料。在社交媒体中,人们使用社交对象作为信息来源。当他们选择信息源时,他们也会过滤信息。此外,社交媒体在一定程度上加强了人们的差异化。由于受到社交圈和自身立场态度的影响,人们往往依附于符合自己喜好的信息圈和意见圈,各种圈子相互孤立甚至对立。
[En]
In modern society, due to the application of the Internet and the development of social media, this phenomenon is even more profound, because some commercial websites provide website materials of a similar nature according to the records of search results. In social media, people use social objects as sources of information. When they choose the source of information, they also filter the information. In addition, social media has strengthened the differentiation of people to some extent. Due to the influence of social circles and their own positions and attitudes, people often cling to the circle of information and opinions in line with their preferences, and all kinds of circles are isolated or even opposed to each other.
在社交媒体中更容易形成的"回声室效应"事实上已经影响到了群体的决策。以至于《科学》杂志最近发表社论表示,社交媒体算法可能对科学传播造成破坏。他们写道:"科学话语的规则以及对证据的系统、客观和透明的评估,与大多数社交媒体上的辩论完全不同。通过用户的愤怒和分歧获利的社交媒体平台,是不是说服持怀疑态度的公众相信关于气候变化或疫苗的科学解决方案的最有效渠道,这是值得商榷的。"
大数据、推荐算法和"过滤气泡"
大数据是近年来的一项伟大发明。互联网平台企业通过各种数据采集工具实现了海量数据的汇聚,相继形成了一个个数据湖、数据仓库、数据集市。
[En]
Big data is a great invention in recent years. Internet platform enterprises have realized the aggregation of large amounts of data through various data collection tools, and formed one "data lake", "data warehouse" and "data Mart" one after another.
在这些陌生而熟悉的概念中,每个个体早就失去了自己独特的棱角,被抽象地建模为一个单调无意义的统计样本,在推荐算法的帮助下,每个用户都不再拥有真正的感知属性,成为冷冰冰的输入工具。
[En]
Among these strange and familiar concepts, each individual has long lost its own unique edges and corners, and has been abstractly modeled as a monotonous and meaningless statistical sample, and with the help of the recommendation algorithm, each user no longer has real perceptual attributes and becomes a cold input tool.
每当我们打开一些带有推荐算法的智能平台应用程序时,这些平台应用程序总是可以捕获我们有意无意触发的一些操作,并向我们推荐似乎满足我们“潜意识”需求的信息。
[En]
Whenever we open some intelligent platform applications with recommendation algorithms, the platform applications can always capture some operations that we trigger intentionally or unintentionally, and recommend us information that seems to meet our "subconscious" needs.
诚然,有时沉浸在信息海洋中的我们,每天都会被各种信息的洪流所困扰,而海量的信息会让我们迷失方向。平台中嵌入的推荐算法使我们能够以更快的效率获得更直接的信息。它让我们受益匪浅,但算法只能根据我们过去的搜索历史,过滤出与我们的观点不同或我们不喜欢的信息,并提供我们想要看到的东西。结果造成了人们的认知隔离,我们陷入了“过滤泡沫”。
[En]
Admittedly, sometimes we, who are immersed in the ocean of information, are plagued by a flood of all kinds of information every day, and the huge amount of information will make us lose our course. The recommendation algorithm embedded in the platform enables us to obtain more direct information with faster efficiency. It benefits us a lot, but the algorithm can only be mindless based on our past search history, filter out information that is different from our views or we don't like, and provide what we want to see. As a result, people's cognitive isolation is caused, and we fall into the "filter bubble".
由于推荐算法刻意迎合大脑的兴奋,使得我们在阅读推荐资讯过程中时候形成"愉悦"的感觉,进而可能成瘾。虽然有时候我们主观上能够感觉到这种成瘾影响了我们的生活,并试图从中逃离,但只要不卸载某些app,又总会陷入这些App的控制之中。
最麻烦的是,目前几乎大部分资讯类app都将推荐算法作为其主要功能,即便是最近我国刚刚出台了相关制度,你也无法逃离其影响。
结语
由于技术发展日新月异,开发者总是要时不时地更新技术,他们自然会被热门的新技术所吸引。
[En]
Due to the rapid development of technology, developers always have to refresh the technology from time to time, and they will naturally be attracted by hot new technologies.
而当某些新的技术出现时,其有时以"革命者"自居,称自己为"架构的天然演进""技术发展的未来趋势",而媒体的推波助澜,技术圈领袖们的故弄玄虚,一方面可能激起开发者的探索欲望,一方面也加重了许多开发者的某些焦虑情绪。
当然,有时使用某些技术可以解决我们的一些痛点,但并不总是这样,这取决于实际的应用场景。如果开发者只基于有限的信息获取来理解他们的环境,盲目地专注于热门技术,他们可能已经陷入了“货物崇拜”(或“货物崇拜”)。
[En]
Of course, sometimes the use of certain technologies can solve some of our pain points, but not always, depending on the actual application scenario. If developers understand their environment only based on limited access to information and blindly focus on hot technologies, they may have fallen into "cargo worship" (or "cargo worship").
我们可能需要更理性地控制自己的信念和结论,运用批判性思维、归纳、数字等更科学的方法,以现实的态度选择技术,把更多的精力放在实现自己的价值上。
[En]
We may need to control our beliefs and conclusions more rationally, and use more scientific methods such as critical thinking, induction, numbers, etc., to choose technology with a realistic attitude and focus more energy on the realization of our own value.
参考链接:
http://www.woshipm.com/it/3513779.html
https://news.cnblogs.com/n/712722/
https://news.cnblogs.com/n/712629
https://wiki.mbalib.com/wiki/%E8%BF%87%E6%BB%A4%E6%B0%94%E6%B3%A1
Original: https://blog.csdn.net/farway000/article/details/123304190
Author: 溪源More
Title: 浅议开发者面临的信息偏差影响因素

Tensorflow lite在Android端部署

语音识别技术论文阅读笔记

【CIPS 2016】(18-20章)多模态信息处理、医疗健康信息处理、少数民族语言文字信息处理 (研究进展、现状&趋势)

深度学习调参trick 调参技巧

指挥调度中心如何选择调度台?

STM32之DAC音频播放

Ad-hoc Video Search(AVS跨模态视频检索)

YARN资源调度过程

pyqt使用graphicsView显示图片

2021/9/8

TensorFlow基本知识及常用函数

论文阅读:An Empirical Study of Training End-to-End Vision-and-Language Transformers

论文笔记:TABERT: Pretraining for Joint Understanding of Textual and Tabular Data

【自然语言处理】【聚类】基于对比学习的聚类算法SCCL
