I don’t think any of your examples are the same, and I would argue that the floppy disk is actually less important than any of your given examples. Each case needs to be judged on its individual merits.
Floppy disks aren’t used any more, are no longer useful, and are no longer seen. As, I said in the post above, they belong in the trash, only to be seen in museums.
If we can redesign things to be objectively better, then change them.
Landline, phone box/booth, and DECT handsets are still visible. They haven’t disappeared yet. But, indeed though, I would also re-design the “Phone”/“Call” icon on modern handsets too – a more human-centric pair of lips with a sound wave coming out would be a better icon for a voice call (sound wave for a voice call, speech bubble with lines for typed text messaging). For example:
Voice call:
Text message:
Envelopes will exist and be used for many centuries to come. Young people might not have sent one, but I doubt anyone in the developed world hasn’t received or seen a letter in an envelope.
Clocks still exist and are seen everywhere. I actually also do find clocks with hands slower to read than digital, and I actually do happen to also have a design of a hybrid clock/watch that is easier to read for a modern audience. That is a whole other discussion for another day though…
English language QWERTY keyboards are indeed a difficult one to change because keyboard usage relies far heavier on trained muscle memory more than any of the other examples. If you are going to redesign the keyboard layout then it would need very strong evidence that the alternative is significantly better. Some people say Dvorak etc. is faster, but I think if you were designing this properly you would need to do a lot of scientific analysis, perhaps including AI tools, to analyse the language and prove that a new layout were significantly more optimized than QWERTY.
A higher priority prerequisite to optimising QWERTY would actually be first to globally standardise the English language itself to be better (other languages like French or German have continuously revised official standards). People are now trying to use English language to write software using AI, and I can foresee this creating many problems due to English not being precise enough. A properly revised standard would have very precise grammar, next to no homonyms and be as near phonetic as possible. Americans like Webster made less than a half-baked go at this, and its half-baked nature just created a mess of different dialects without global consensus and little benefit [insert infamous xkcd comic here
]. It would need decades of consultation, global consensus building and AI analysis to improve and standardise English properly though! Webster couldn’t even get “color” right - it should be “culor” or “kulor”! “Center” I sort of agree with, but might have been better as “senter”. Anyway, this is a long separate discussion that is getting off-topic…