在IETF，如何推进自己的想法？

针对人们在参与互联网标准制定中存在的一些问题和误区，日前，IETF的资深研究人员Nick发了一份邮件。

邮件是发给一个名叫Khaled的人。Khaled之前提交了自己的一个新的协议方案草案，自我感觉非常好。但该方案在IETF并没有得到采纳，他感到很生气。这一过程持续了两年，他一直说自己的方案非常好，但自己不是程序员，无法编程验证。

最近他又换了一个工作组，重新提出他的草案，并发出关于IETF不够重视其提议的公开邮件。

针对此，Nick提出四点建议，这些建议提到的一些问题对于想在IETF提交草案并期望能够成为RFC标准的研究人员同样具有参考价值。

邮件内容如下

Khaled：

过去几年中，有很多人看过你提议，他们通过几百封邮件的讨论交流后一致得出相同的结论：你的提议行不通。实际上，这也意味着，你是在要求IETF工作组处理一个他们觉得行不通的提议。

如果你想让IETF慎重考虑你的想法，那么你首先需要证明这些提议是可以实现的。那你就需要从倾听和处理意见开始，尤其是那些被多次提出着重讨论的问题。

关于此，我有几个建议：

1. 写一份你所提出的提议是如何工作的原理实现。或者是讲清楚，你所提出的技术是如何与IPv4或者IPv6网络建立连接的？

2. 更新其他协议的规范文件以支持你的提议，如路由协议：BGP、mpls、OSPFv2、OSPFv3、ISIS等。仅针对这些协议就至少有500个RFC，所以为什么不选择一小部分进行更新，使其能够支持你的想法？如果你能编写出一个有效的实现文档，应该会更好。

3. 为主机应用程序编写一个API规范以解决双重寻址问题。

4. 写一份你所提议的“路由协议”的实现细则，它应允许一个网络与另一个网络交换路由信息。专业提示：确保它能在你提议的技术上工作。

说你不是程序员以及让别人为你的想法编写代码必然是不可取的。现在的问题是，许多研究人员已经明确表示你的想法不可行，如果你希望你的想法被认真、慎重对待，那么你有责任去证明他们的想法是错误的。

一直争辩别人应该认真对待你的想法，这件事也是没有用的。除非你能证明它们是可以工作的，是确实有效的，否则人们不会认真对待它们。

当你写出代码证明你的想法确实可行，然后再回到IETF，也许那时人们会更认真地对待你的想法。

Nick

点此看英文信

To Khaled

Khaled,

There have been hundreds of emails over the last couple of years written by people who've looked at your proposals and who have unanimously come to the same conclusion: 

None of your proposals are going to work.

So when you ask an IETF working group to work on your proposals, you're asking people to spend time on something they think is not going to work.

If you want the IETF to take your ideas seriously, then you need to prove to people that they're going to work.

A good way to start would be to listen to what people are saying, and fix some of the more serious problems that people have patiently taken time to comment on.

Here are some ideas:

1. write a working implementation of "ipv10".  Or even a proof of concept which allows a single application to make a single connection over "ipv10" to another ipv4 or ipv6 host on another network.

2. update the specifications for other protocols to support "ipv10", e.g. routing protocols: BGP, mpls, OSPFv2, OSPFv3, ISIS, etc.  There are at least 500 RFCs for these protocols alone, so why not take just a small selection and update them to support "ipv10".  If you wrote a working implementation, then that would be even better.

3. write an API specification for host applications to handle dual addressing.

4. write an implementation of your "routing protocol" which allows one network to exchange routing information with another network.  Pro tip: make sure it works over "ipv10".

It's no use to say that you're not a programmer and that it's someone else's job to write the code for your ideas. People have made it clear that they think your ideas won't work, so if you want your ideas to be taken seriously, then it's your responsibility to prove them wrong.

It's also useless to argue endlessly that people should take your ideas seriously.  People won't take them seriously unless you can show that they actually work in real life.

When you've written code to show that your ideas work then come back to the IETF and maybe at that stage people will take your ideas more seriously.

Nick

  来源：中国教育和科研计算机网

往期推荐