Philosophies, quotes and mottos from the professor. No need to provide every sentence exactly the same. Suggest to put similar things together without chronological order.

THE READER ACCEPTS ALL STATEMENTS PROVIDED AS IS, WITHOUT ANY WARRANTY; WITHOUT EVEN THE IMPLICATION THAT EITHER THE AUTHORS OR THE PROFESSOR HAVE SAID, KNOW, SHOULD KNOW, OR AGREE WITH ANY WORDS PROVIDED.

• When people collaborate, there will be conflicts.
• CS is not physics. There is no Bible telling people what to do. All designs face TRADE-OFFs.
• People write articles to express what lessons they learned. Not only results but the insights why we get results should be included.
• (About ALOHAnet) Simple systems used to be robust.
• (About OSI) Trials & Errors vs Thought Design.
• (About hICN) People tend to use mature systems. And you cannot ask all end-users speak your protocol. Thus, the battlefield of NDN is edge computing, not CDN.
• (About ndn-repo) Do not generalize. We shouldn't invent requirements.
• (About subnet) What's changed is not the format but interpreting, given the fact all devices had implemented the format.
• (About IPv6) The market is not going to spend a single penny when the existing system works.
• I know Sunday is Sunday, but graduate students work on Sunday.
• You cannot simply say the compatibility is important without inspecting why.
• Proof of authority only proves you are you, without proving that you are honest.
• There are always intuitions in protocol design, but not proofs.
• Do not use any hardware-based examples in protocol specification, because hardware changes too fast.
• (Self Interests in blockchain)
  • In peer-to-peer scenario, there is always a question to be settled: "Why should I help you?"
  • If I have to do this, what's the benefit?
• Writing down a problem in half of a paper. You write down the difficulty and people will pay attention to that. The industry people need to know what need to be improved.
• Don't plan to do writing later. Later your thought is going to fade.
• Congestion is something abnormal. That's why I dislike congestion control fairness. The network is not always congeted and the complex mechanism itself becomes a factor of attack.
• (About security) Whatever protocol you use, you must start with something shared.
• (Junxiao said that) Sleep is No.1, I agree.
• (About paper writing) When you don't have time to write a short one, write a long one.
• There are two kinds of constraints: physical limitation and artifacts of today's product. Point out which.
• (Routing protocol)
  • Routing & forwarding is the essence of networking - how we deliver from anyone to anyone.
  • Every node is blind. They can't see anything. That's why they always chat.
• An address is a special case of a name only given to interfaces. A name is its generalization which is given to everything.
• Slow start is a golden rule for everything.
• (Application driven)
  • If you only write papers, you won't know whether it actually works.
  • Academic guys do systematic studies and operators solve problems (in a quick and dirty way), so there are so many "Operators Hack" today.
• NDN doesn't promise security, but it offers blocks to build a secure system.
• Human by its nature is overly generalized, not good for system design.
• It's the end-users but not big companies who control the future. That's why Internet and Ethernet are not killed.
• Read books. You really should learn knowledge from the history.
• Paper is a byproduct.
• If you want to convince the reviewers, convince yourself first.
• Write a paper in the eyes of readers. Don't just say what you want to talk.
• (HOSTS.TXT) One thing human definitely do is making errors. So there's a very important problem: who's in charge of what.
• For each successful thing, know what factors lead to its success.
• Performance is the result of operations, not design.
  • How could you measure the speed when you are designing the upper protocol, without even knowing the connection under it?
• But I'll tell you the whole Internet lives on our own money. The aggregation doesn't work that much today.
• (About DNSSEC)
  • You should ask "Where things go wrong?"
  • The real world is not simply Alice talking to Bob.
  • In that days people know nothing about "immutable data".
• Privacy is a losing battle.
• (DDoS) The problem is it's getting worse with time, and we don't have solutions better than 20 years ago.
• You must not depend on lower layer's error reports for high-layer normal functioning, because it's optional.
• Devil hides in details.
• Nothing is for free. How they make money from you?
• The basic NDN idea is Data = Name + Content + Signature.
• (Packet switching) Reliable system should be made out of unreliable pieces.
• (RIP vs Baran's FIB) All routing protocols do flooding, so the data don't need to do so.
• Whatever artifacts designed by human not only can be changed, but will be changed. This is because you design something at a given time, based on what you have. But technology advances.
• DNS is not about trust, it's about uniqueness.
• (Babysitting) Leave it there. Spoiling is the real problem.
• It's not who's doing better. It's about the insights: interpreting the results.
• Think of best solution we could have, not what we have now.
• If there are multiple solutions, finally only one will be chosen.
• Routers do multicast, so user data do not need to.
• The only methods to provide reliability are repeating, acknowledge, both.
• Do meaningful work. Don't just write for reviewers.
• Let's forget the result from what have discussed last time or what we have implemented. We need to focus what is right.