> Then we got MODs and Midi , and suddenly we had MP3s clogging the existing 56kbps infrastructure.
This is just wrong and not in any way analogous to some blockchain network getting clogged up.
Your first problem is you've got your causation backwards. Fast dial-up facilitated access to larger media files. People didn't get 56k modems to download MP3s, they downloaded MP3s because they had 56k modems. Even then MP3s (and other media) were commonly 64k and 96k rips because people had limited bandwidth and storage space. You'd only find 128k MP3s and the like on specialty F-serves and FTPs.
Your second problem is RealMedia content was very specifically tuned to work over dial-up connections. RealVideo's buffering problems were very often last-mile bandwidth and latency issues. Even great 56k dial-up had quarter second RTTs, single digit percentage packet loss, and topped out at ~45kbps actual payload throughput. RealMedia servers (and the file and stream formats) explicitly supported multiple streams based on your available bandwidth. That's why RealPlayer had a dropdown asking what type of Internet connection you had, a single RM file or stream endpoint could actually serve multiple streams at different bitrates just like today's HTTP streaming. While a single RealMedia server might be congested if it had tons of clients the backbone networks had plenty of bandwidth and were typically not "clogged up" from RealMedia servers.
Lastly, the problem with Ethereum and other blockchains is their backbones are clogged up with speculative transactions. They're not scaling to meet current demands for transactions and only maybe possibly will scale in the future to handle future amounts of transactions. The Ethereum case is in fact the clogged backbone you were trying to ascribe to the Internet of the past.
