I still use Vim as an editor almost daily without really thinking about it, and think that the modern reimplementations such as NeoVim (serious), Vis (a model editor with multiple cursors? Intriguing!), & PyVim (a Vi clone implemented in Python? That seems like madness!), are worth exploring either for actual use or as hobby projects. I have tried to learn to like Emacs, but I do not.
If people would stop sending me html-formatted email, I would probably still be using mutt as an email client, but that ship has sailed. No one has yet made a good solution for managing calendars. I agree top-posting is illogical, but email-client design is generally so terrible that I don't think we can fix that habit in most contexts.
Usenet, IRC, XMPP, and Mailing Lists were all excellent technologies for managing conversations of different kinds online, but I suppose that it was inevitable that eventually The September that Never Ended and corporate greed would bring us web applications that would manage global communications terribly badly while trying to lock users into walled gardens, and users that accepted this state of affairs. Sometimes I have been given hope that something better might emerge, only to have it dashed. Google Wave was a very good idea with a terrible security model.
Younger me loved learning how Linux works. Older me needs a computer that works reliably but with many of the same tools as a *NIX platform and yet the ability to run Microsoft Office, and so older me accepts Mac OS as an acceptable compromise. In general, please don't ask me to pick a side in the platform wars. I will work with anything but I will not love it; I am still mourning the loss of RISC OS.
Open Source software is an excellent idea, but many of the advantages are overstated for all but a very small number of projects, and I am also happy to pay for software. I wish that the OSS had not become a model for academic publishing, and I wish that the people trying to push that agenda would stop. When I pick an OSS licence for my own hobby code, I usually choose the BSD one.
I still write any important piece of academic writing in LaTeX first, but I wish that the tools for scholars in the humanities were as good as those for scientists. The best Desktop Publisher ever was !Impression Publisher on RISC OS. I find it astonishing that it has never had an equal on OS X, Windows, or Linux.
Yes, using PGP/GnuPG is a very good idea, and I still know where my key is. As soon as the standards for ECC cryptography have settled down, I'll probably make a new one. The fact that people do not use PKC for their communications raises all kinds of interesting issues around user-interface design that are not unrelated to my academic interests. Teaching people to use unfamiliar software, built upon unfamiliar technology, interacting via unfamiliar and imperfect metaphors is difficult, especially when no one has read a software manual since around 1990.
Of course the internet allows censorship, but I do remember the days when people would say, 'The Net interprets censorship as damage and routes around it', and really believe it was true. It was a more optimistic time. Since then, we have disproved the idea that what was holding most of humanity back was simply access to knowledge. It turns out that the internet propagates falsehood and stupidity as well as it transmits anything, and there are infinitely more falsehoods than truths. (I will never get back the time I spent arguing with a philosopher who, having heard me say something along those lines, engaged me in a long debate about whether there were infinite falsehoods for every truth or merely uncountably many. He had the philosopher's special talent for missing the point I was making and engaging with an entirely different one.) The research papers into the human willingness to believe falsehood rather than truth are depressing reading.
The people who wanted computers to adopt a Spacial User Interface were correct. It is a shame that they lost that battle, because they were correct in their understanding of the way that human memory operates.
If the earliest computer spell-check systems had provided two UK dictionaries rather than just one, people would still be (correctly) using '-ize' in the majority British academic prose and perhaps even in serious newspapers. We might even still be writing 'connexion', too.
If I ever find the time, I am sure that I shall write about some of these on my blog.