A decade and change later when Intel started asking for hardware copyright[1], Congress actually realized their mistake and created a sui generis right for integrated circuit designs ("maskwork rights"). We never complain about maskwork cases because they are incredibly limited, have a 10 year lifespan, and basically only cover direct copying. Software would have been no different had Congress not made the mistake of handing the full enchilada of copyright monopoly protection - including all those pesky questions about derivative works - to software companies.
[0] e.g. is linking infringement? can you copyright interfaces?
[1] Fun fact: before this, it was entirely legal to X-ray chips and reproduce them. In fact, the NES famously shipped with a bootleg 6502 that only had decimal mode missing because that was the only thing MOS could patent.
The lack of legal foundation for CDL is not entirely an accident, however. First sale and similar copyright exhaustion doctrines are hard-fought and won rights of the reader. But these rights rely on the fact that no actual copying is taken place. When you interact with any copyrighted work using a computer, there is an almost gratuitous amount of copying going on. If you so much as cough on the work, you are breaching copyright. Publishers know this, and they have been very successful ramming "licensed and not sold" language through the court system.
The funny thing is, while pirates have been stereotyped as waving their hands in the air and shouting "technology" to opt out of the law, publishers have been way more successful at doing the same, even though they fought tooth and nail against digital distribution.
[0] In the ReDigi case the "digital resale" software was even specifically engineered to erase parts of the file as they were sent to the new owner so that the number of duplicate bits floating around would be negligible at any particular time.