At the meta level, my life is already too short to allow closed-source tools for anything on which I may need to depend in the long term. Commercial entities come and go, for reasons important for them but not necessarily for me. Old, ancient code can be made to run, as long as it's available and legal to distribute and run. If it's useful enough, I will be not the only user by far, and there'll be enough collective…
On the inverse, even as a FOSS advocate, my life is too short to only use open source software all the time and to reinvent the wheel and tools that have been put together into one fantastic product by the combined effort of hundreds of people. I used Joplin for years, I even self hosted the Joplin Sync server, but Obsidian + Community addons runs literal miles around it from a performance and user cusomizability sta…
My point is that for long-term, high-investment stuff, in my particular view, the possibility of interruption and loss outweighs any possible upside, and any migration would be costly. Similarly, a move from a house where you lived for 20-30 years by its disruptiveness may tantamount to surviving a flood or a fire.
For "short-term" tools this is not so, very obviously, but they do not need my investment, hence they would incur no loss were I to migrate off them if a replacement exists. I could not replace Google with anything, but then DDG and then Kagi appeared, and largely replaced Google web search for me, quite seamlessly. When a better LLM appears, I will start using it instead of Claude. Etc.