德莫斯研究所-良好的基础：为什么民主国家应该关心互联网的连接（英文）-2021.3正式版.doc

《德莫斯研究所-良好的基础：为什么民主国家应该关心互联网的连接（英文）-2021.3正式版.doc》由会员分享，可在线阅读，更多相关《德莫斯研究所-良好的基础：为什么民主国家应该关心互联网的连接（英文）-2021.3正式版.doc（22页珍藏版）》请在淘文阁 - 分享文档赚钱的网站上搜索。

1、GOODFUNDATIONSWHY DEMOCRACIES SHOULD CARE ABOUT THE WIRING OF THE INTERNETJOSH SMITHCIARAN CUMMINSALEX KRASODOMSKI-JONESMARCH 2021Open Access. Some rights reserved.Open Access. Some rights reserved. As thepublisher of this work, Demos wants to encouragethe circulation of our work as widely as possib

2、lewhile retaining the copyright. We therefore have anopen access policy which enables anyone to accessour content online without charge. Anyone candownload, save, perform or distribute thiswork in any format, including translation, withoutwritten permission. This is subject to the termsof the Creati

3、ve Commons By Share Alike licence.The main conditions are: Demos and the author(s) are credited including our web address www.demos.co.uk If you use our work, you share the results under a similar licenceA full copy of the licence can be found athttps:/creativecommons.org/licences/by-sa/3.0/legalcod

4、eYou are welcome to ask for permission to use thiswork for purposes other than those covered by thelicence. Demos gratefully acknowledges the workof Creative Commons in inspiring our approach tocopyright. To find out more go towww.creativecommons.orgThis project was supported by:Published by Demos M

5、arch 2021 Demos. Some rights reserved.15 Whitehall, London, SW1A 2DDT: 020 3878 3955hellodemos.co.ukwww.demos.co.uk2CNTENTSEXECUTIVE SUMMARYPAGE 4INTRODUCTIONPAGE 5WHAT ARE PROTOCOLS?PAGE 6WHO GOVERNS PROTOCOLS?PAGE 11RECOMMENDATIONSPAGE 173EXECUTIVE SUMMARYThe design and policing of online spaces b

6、ears directly on many of the most important problems faced by democracies today. Many proposed solutions, however, barely skim the surface of the Internet as a technology. Unseen beneath growing public debates around privacy, misinformation and online harms lie layer upon layer of infrastructure and

7、 code, sending the messages which make up the visible parts of the Internet. These layers are governed by technical protocols, changes to which affect the complex human societies which exist online. Those aiming to build a better web cannot afford to ignore the foundations upon which it is built.Pro

8、tocols, and the Internet more broadly, are developed and governed by a handful of acronym-friendly groups - the IETF, the ITU, the IGF - among others. Many operate on the principlesof multistakeholderism - a form of democratic governance which helps groups of organisations to work on common problems

9、, spreading power and decision-making out across stakeholders. This ideal is worth protecting, and it will require concerted effort from their members to do so - as well as to defend these groups from capture by the powerful, and from closing themselves off to outside ideas.Technology companies and

10、states both wield outsized influence over the Internet. Their power is exerted both over the organisations which develop protocols, and over how and whether their recommendations are implemented. Those able tocontrol a countrys communications infrastructure, or change the settings in web browsers us

11、ed by millions, can unilaterally affect the protocols used by customers and citizens, changing the types of data sent by their computers and the actors to which that data is sent - often without their knowledge.In this paper, we examine two examples of protocols which may represent the next evolutio

12、n of the Internet: New IP, developed by Chinas Huawei, and DNS over HTTPS (or DoH), developedand championed, among others, by the Mozilla Foundation and Cloudflare. For each, we discuss the significant effects which these changes could have on the human and social layers of the Internet. We also tak

13、e a hard look at the multistakeholderist groups shaping the rules of the Internet.The relationships between governments, civil society, the groups who govern and develop protocols, and the technology companies who often employ their members are critical in the battle to builda better web. Where moti

14、ves can be aligned, alignment should be found. Where they cannot, those without the power to implement protocols must be empowered to interpret and challenge the decisions of those who can. To do this, it is crucial that policymakers, citizen groups and technologists are able to find common ground.

15、With this paper, we hope to contribute to this.4INTRODUCTIONHOW CAN STATES AND CIVIL SOCIETY MAKE THE INTERNET BETTER?Across the Western world, the period of unfettered growth and development online is coming to an end as governments begin to roll out legal and regulatory regimes. These changes, in

16、the eyes of regulators, legislators, the media and the public, have been spurred by scandal. People from across these groups looked at the Internet, and decided they werent happy with what they saw.The public attention paid to the Internets content layer - the thin, visible crust that sits on top of

17、 dozens of layers of technology and infrastructure - might well be distracting us from where change actually takes place. What if the battles for a good web, fought at present across those parts of the web we can see, are doomed to be lost on the bits we cant?This is not to say that those arguing fo

18、r a new settlement over the web have been fighting the wrong battle. The content layer - Facebook, WhatsApp, Google, 4Chan and the billions of online spaces and applications that comprise it - is worthy of attention. Decisions made around how the Internet allows people to represent themselves and ex

19、change information affects the lives of billions. But we must not lose sight of the foundational layers that make up the Internet. These layers are governed by a set of standards far more precise and effective than we have been able to apply at the content level: Internet protocols - a set of conven

20、tions governing the transmission of data.These protocols detail best practice for those who run the infrastructure underpinning the web, and are under constant review. There are a number of reasons why they might change. Some changes are motivated by a wish to solve difficult problems, by engineers

21、optimising for faster or more stable data transfer. Other changes have been made bycorporations interested in improving their services to boost profits or consolidate market power, or states wishing to control the information accessible to their citizens. As we will see, changes propelled by these m

22、otives are often presented as a way to increase speed, improve the user experience, or to make people safer.While many of these proposed changes are discussed the public domain, many are couched in impenetrably technical language and inscrutable to those affected. Representatives tasked with navigat

23、ing the social impacts caused by these changes - let alone end users - are too often unable to judge what the impact of new proposals might be; to see how they will affect the balance of online power between citizens, governments and technology companies. This must change. Just as policymakers, the

24、public and civil society should care about how these spaces are governed at the content level, those who want to protect or improve the web need to be able to go deeper.This paper explains what protocols are: how they work, why they matter and how they are governed. It explores the implications of t

25、his approach, and urges a closer relationship between the architects of Internet protocols and those responsible for the publics they affect. This is not to say the public need to understand the intricacies of routing protocols or HTTPS - indeed, the web has to a great extent succeeded because it al

26、lows the uninterested not to care about these things. However, we need to get better at discussing the social impact of technical changes. To do this, those who want to solve human problems on the Internet need to account for the changing technology which underpins it, and those who are interested i

27、n the technology cannot ignore the human impact of their decisions.5WHAT ARE PROTOCOLS?In order to make the case for the importance of protocols to online life, it helps to understand a few long-standing architectural features of the Internet. Two of these are the Internets layered design, and the e

28、nd-to-end principle, each of which we touch on below. These are not full technical descriptions, but aim to provide some fundamental points.INDEPENDENCE BY DESIGN - THE LAYERS OF THE INTERNETThe Internet is made up of separate but interconnected layers of technology, each of which can be changed wit

29、hout impacting the rest of the stack.The Internet can be separated into distinct conceptual layers, each of which does a different job. At the top sits the application layer. This is the visible, functional part of the Internet, composed of programs which fetch and display videos and email, control

30、what gets onto your Facebook feed and conduct the million other tasks to which online data can be turned. At the bottom is the layer of physical matter which carries Internet traffic - miles of copper wire, glass fibre and air that carries wireless transmissions.There are a few models for what happe

31、ns between these upper and lower layers - how the electrical impulses carried by wires are translated to applications. Some models have seven layers in total, others four - it depends whos telling the story. For our purposes, the detail here is besides the point, and for our purposes we will refer t

32、o this middle section as the transport layer.1 Its job is to makesure the data which makes up the application layer gets to the right place, and can be understood when they arrive.Crucial to the success of the Internet has been the fact that these layers are designed to talk to each other through st

33、andardised interfaces, but otherwise work independently. If I am writing code on the application layer - designing, say, a website which allows people to share videos - I dont need to know whether my users will access my site by copper wire or through WiFi. Their videos will upload either way.This l

34、ayered design has significant benefits. It has allowed a diverse group of people to improve the Internet by working on the problems which hold their interest - whether thats designing accessible websites or the best way of shining a laser down a hair-like strand of glass - without their choices and

35、innovations disrupting the layers above them. The Internet owes much to the work of engineers and other interested parties informally repurposing existing systems for new uses, and contributing new code where functionality was missing.The problem with this neat technical separation lies in the layer

36、 above the applications: it lies with us. The very topmost layer of the Internet, of any technology, is unruly, chaotic, and refuses to be confined: unlike tweaks to the efficiency of network cables, problems in the social and content layers - the human layers - influence, and are affected by, every

37、thing else in the stack.1 For a deep drive into one of these models, chapter 4 of Jonathan Zittrains book The Future of the Internet (And How to Stop it) provides a good overview. See Zittrain, J. The Future Of The Internet And How To Stop It. Yale University Press, 2008. Available at http:/yupnet.o

38、rg/ zittrain/ accessed 15 March 20216EFFORT AT THE EDGESBy design, processing happens at each end of Internet messaging, rather than in the middle. This makes centralised control and oversight more difficult.Many of the protocols which make up the transport layer of the Internet have stood the test

39、of time. TCP/IP - a pair of protocols which, in a nutshell, ensure that the messages which make up the Internet can be i) properly addressed on sending andii) assembled and understood once theyve reached their destination - was first proposed back in 1974.While tweaks and improvements have been made

40、 to these protocols, their fundamentals have remained unchanged since their inception. Together, they still underpin the majority of Internet traffic: in 2010, 85% of Internet communication took place over TCP/IP.2 However, as we explore below, new protocols are arising to take their place.The flexi

41、bility of these protocols has enabled the Internet to keep delivering packets through an explosive expansion of users - and uses - over the last 50 years. It has also influenced the ways in which the network is designed and used. One impactful decision made by Vint Cerf and Bob Kahn, the engineers w

42、ho first proposed the protocol, is the much-cited end-to-end principle. This essentially states that the difficult tasks involved in packaging up and addressing a message, and in processing them once theyve arrived (along with other services) should happen at each end of the network, rather than bei

43、ng the job of the machines which route the messages. In Cerfs design, the Internet has smart edges - computers, phones and servers - and a simple middle - routers whose only job is to pass messages efficiently from one point to another.This design makes the Internet more difficult - though not impos

44、sible - to control centrally, at least at the network level. (Recent events such as the effective removal of the social media site Parler from the web suggest that other forms of central control - especially from groups of companies with effective monopolies over distributed web hosting - are still

45、very much possible.) Since TCP/IP packets are designed to be routed in the same way whatever data they contain, and dont require any knowledge of who is at each end of the packagebeyond an IP address, its theoretically impossible for systems using this protocol to treat different types of communicat

46、ions differently - for instance, for Internet service providers (ISPs) to charge customers more for gaming than for accessing Wikipedia.THE EROSION OF END-TO-ENDOver time, states and corporations have found ways to better oversee what information is being shared over the Internet, but these techniqu

47、es are implemented in spite of the protocols as they were designed.That, at least, is the theory. In practice, those machines in the middle of the network - including ISPs, corporate networks, and the companies who build applications - have developed a variety of methods to filter, censor and shape the traffic which passes through them. As the Internet Society pointed out in 2004, these shifts have often been prompted by external forces, as “corporate network administrators and governmental official