January 1, 2025　　# New Year Greeting

Wishing everyone a happy and prosperous 2025

January 2, 2025　　# On Moore's Law

It was around the beginning of the 21st century that I began to enjoy the financial freedom of buying PCs of my choice, and Moore's law was a blessing in that matter. Although I never bought any cutting-edge PCs exactly because of Moore's law, which means a cutting-edge PC is merely a year ahead of a PC at one-third the price, and I never felt the desire to triple my spending for that. As time went by, however, Moore's law, though hasn't been declared over, has gradually diminished from a 1-year doubling time to 18 months, and then 2 years... Recently, I bought a new external hard drive and noticed in my purchase history that I had bought a 2TB external hard drive in 2019 at $59.99. More than 5 years later, as we are now, 2TB remains a common option for an external hard drive, and with the same price, one can, unlike in the era when Moore's law was at its full power, hardly buy anything better. Moore's law in the area of external hard drives seems practically over. [Added Notes (in reply to a comment): Sorry for not being clear about the difference between Moore's law and other technological trends that are sometimes called "Moore's law of ..." or, as Wikipedia put it, "strongly linked to Moore's law". I always try to limit each topic to a thread of no more than 2 posts (thus no more than 1,000 characters), and often end up cutting corners because of that. 😅]

January 9, 2025　　# "Don't forget to like this video"?

Is it just me, that feel odd to see the phrase "Don't forget to click the subscribe button and like this video" commonly used by YouTubers? One can't "forget" or "remember" to "like" something. One way to make sense of it is to extend the meaning of "like" to mean "click the like button", but then "click the subscribe button" becomes an ironic redundancy since "subscribe" alone would be sufficient. It's far better and much more consistent to say "Don't forget to click the like button and subscribe to this channel". :-)

January 16, 2025　　# Los Angeles Area Wildfires

The wind-driven Los Angeles area wildfires have destroyed thousands of structures. From time to time, however, a house almost undamaged can be seen surrounded by numerous structures that were burned down. Some news reporters attributed that to a luck factor: wind-driven wildfires sometimes skip a house. While luck factor may occasionally play a role, it is far more likely that fire-resilient design is both the true savior and the major controllable hope for the future. The reason wind-driven wildfires are particularly devastating is because embers can fly faster and further. For a given house, in the typical duration it takes for its neighborhood to burn down, the chance that no ember happens to hit it is extremely small. On the other hand, as long as the roof and outer surface of the house can sustain the much shorter duration of the embers burning here and there without being ignited, the house will most likely survive.

It is unfortunate, though, that even in areas where wildfires are a known threat, and even among million-dollar properties, only a very small percentage of the structures are sufficiently fire-resilient.

January 17, 2025　　# On Spam

A crude and yet effective criterion of spam: Any email that starts with "Congratulations!" is spam, and so is any letter with "Congratulations!" on the envelope. Spammy as it has become, "Congratulations!" is nevertheless still a popular bait, because people's childish fondness for good surprises is as stubborn as ever, even when the surprises are from unsolicited sources. :-)

January 23, 2025　　# Translation and Copyright

I know asking a question on an account where almost no posts have more than 100 views is naïve to say the least. But writing is more about expressing, and it’s not a big deal to phrase a curiosity without an answer. So here is the question anyway: Suppose a book in language A has passed its copyright period, but a translation of the book in language B is still copyrighted. Now suppose someone tries to translate that book into language C. While translating from the version in language A is fine, translating from the version in language B would be a copyright infringement. But it would be difficult to tell, even more so to prove, which version has been used in the translation. Is it just up to the translator's self-discipline not to translate from the version in language B?

January 30, 2025　　# The Obscurest Epoch Is Today

「History is much decried; it is a tissue of errors, we are told, no doubt correctly; and rival historians ex­pose each other’s blunders with gratification. Yet the worst histo­rian has a clearer view of the period he studies than the best of us can hope to form of that in which we live. The obscurest epoch is today...」

—— Robert Louis Stevenson ("The Day After Tomorrow", 1887)

For however different an era that might be from ours, and regardless of the thoughts originally behind it, those lines are nevertheless quite valid even today ‒ perhaps especially today, when partisanship, irrationality, misinformation, conspiracies... that can obscure people's view are far more corrosive than they ever were, thanks to the technologies that are far more effective in spreading cheap propaganda than rational thought.

February 5, 2025　　# Novelists And AI Training

There was a time, not too long ago as I remember, when a large group of novelists urged AI companies to stop using their works in AI training. I'm not a fan of AI writing, especially in the area of novels, in which the joy of reading is, at least for me, largely because the authors are fellow human beings whose experience, intention, and emotion I feel able to resonate with. That said, I nevertheless don't think novelists had a valid point against AI companies. AI training, so far as I know, is correlation-based and involves an amount of material far larger than any human writer could possibly read. As a result, any particular piece of work used in AI training has negligible influence in forming AI writing style, far less than a book read by a human writer might contribute to his/her writing style. Since we don't consider the latter to be a problem, I see no reason to limit AI training in the way those novelists wished.