Deception often mimicks as explanation. There's not much of "explanatory power" in this analogy. The only real message a was not one of "understanding" but of associating the phenomenal success of TCP/IP to the LN, in the same way that one associates "buying cleaning products" to "be a successful housewife and have great husband and kids" in TV commercials.
This analogy is used correctly to explain that the only way to scale a blockchain is to add an additional layer. Much in the same way as you can increase the bandwidth of an Ethernet segment from 10Mb to 100Mb to 1Gb but it needs routing to making scale globally.
And that's exactly where the "analogy" doesn't apply, as I showed, just to get a reply that "of course, an analogy doesn't fit althe l the aspects".
I cannot tell you how amusing it is that rather than address that you first start by trying to push the analogy past what it intended to explain in order to try and discredit it and now you're trying to discredit the use of analogies altogether. One might come to the conclusion that you're trying to distract attention from the fact it is correct and avoid the debate being on the subject originally intended in this thread.
On the contrary. First of all, there is no scaling problem in the actual structure of bitcoin, because, exactly, it is a client/multi-server system, and not a P2P system. So there is no scaling problem to be solved.
But next, the routing problem in TCP/IP is an entirely different problem than the common consensus problem in bitcoin. TCP/IP is concerned with moving data between users, bitcoin is concerned with everyone agreeing upon a data set.
On top of that, the big success of TCP/IP routing lies in the ease by which connections can be set up and broken. This is what allows routing to be so efficient in TCP/IP. However, in the LN, setting up a link is costly, and stops you from setting up another link, because you lock in funds. In TCP/IP you can set up as many links as you want, and if some break down. that's at almost no cost. In the LN, if your channel is dead, it takes fees and a waiting period to unlock them and to set up another link.
So many of the elements that made the success of TCP/IP are not applicable in this analogy. Both solve different problems to start with, and the great success of TCP/IP comes from the easiness of re-routing, while exactly that is costly and slow on the LN network.
Unless, unless, your LN partner is a very reliable, trustworthy partner, who has very reliable, trustworthy partners and so on.