On the other hand “@” and hashtags “#” are contained in the message body, as are urls, some of which get the equivalent of “deep packet inspection” by Twitter (although more by the end-apps than the “network”). So in these intstances Twitter is an “intelligent” network.

It’s never black and white (even with the Internet, as implemented in practice, as we have discussed many times) but I’d say Twitter leans toward the end-to-end principle.

I see why you would disregard Twitter becoming “intelligent”. Indeed it is not becoming, but it always was. Afterall, if I start my tweet with “D” it assumes that it is supposed to be a DM. In this respect it is not really end-to-end. Twitter server is most definitely in the middle.

I am not being argumentative. My contention is that “Stupid Network” paper is wrong in its claim that PSTN is not a stupid network. After call establishment, the switched network was transparent to the traffic. The only hitch was echo cancelers. Of course, this was circumvented with the modem signal turning them off. Once it was done, the data flowed truly end-to-end, albeit at a slow rate. That is why PSTN could support data/fax modems. Why it even supported devices that used different codecs. The problem David faced was that his project wanted all the phones to experience without requiring any changes to the end devices. In a way his project REQUIRED an intelligent network, but he faced a stupid network.

So I say PSTN is a stupid network and Twitter is NOT.