But it originates with a very real and very damaging operational problem, that of BSD 4.1's predilection to TCP Silly Window Syndrome and an operator's desire to minimize the impact of that on competing data traffic.
Dear Fred,
thank you for your inputs. You point out the IETF problems I see. You say developers and users may meet. And the example you take is an "operator" (your word) predating IETF/IAB because you know it better. This exactly documents the four problems I point out.
1. you describe IETF as only a forum, like "Nature" for the Science community. I would say it is more. It certainly is a place of dialog between techies and operators (I agree with your meaning of the word, John Klensin rebuked, of someone who operates something in using an IETF deliverable).
2. but it is not the polylog (the very nature of the internet) we need, with all the users (name them @large?) content providers, politics, business etc. So there is a split we often observe in here between reality and assumptions. You answered me that there was indeed some experimentation in the Standard track. I responded there is no global societal acceptance (including economic model, general policy etc.) nor governantal (will the governance accept the proposed solutions and will it be able to manage them?).
3. as such IETF it is not a place for major innovations. This calls for architects. The main problem of IETF is IAB and the lack of an Internet global model and of global development targets. You use an example from a time when major changes could be made, just because they were correct. Today the size of the network prevents to do that easily. This is to go be correct direction and momentum.
4. you use examples you know. You document this way a major problem. You should be able to use any subject as an example. A person like you is trained to understand a technical state of the art summary easily in a few hours. But there are none of them. Read all the RFCs and the archive of their closed WGs.
- IETF for geeks and internuts only.
- awkward with users (demands, acceptances and usages brainware)
- no IAB vision based upon a commonly accepted global model
- no state of the art summaries and organized RFC comments and experimentation reports
Experience is the mistakes of someone with memory. Why not to analyze the IDNA (now confirmed failure) and to try to understand how not to repeat it? (I opposed IDNA all the way long, but I am the first to say it was a complex work achieved by a hard working WG). It was useful in uncovering problems to be solved and some solution elements.
jfc