Thoughts
2009-02-15: Genres that are NOT genres
Pop is NOT a music genre. It's a parallel classifier as irrelevant as the color of the album art. Unlike true genres, the definition of 'pop' is subject to the temporal whim of whatever fits on the radio station's charts. Considering that 'pop' twenty years ago was radically different than what is called pop today, I think this pseudo-genre should be forcefully cleansed from all MP3 tags. Admittedly, other genres do gradually shift and subdivide into others, but at least they are based on something meaningful, like instruments used, rhythm, and other song attributes.
The second most amusing 'genre' is electronic (note I don't mean the more specific EDM). Really, this is quite an irrelevant one, considering it is merely the method of audio production and doesn't really tell anything more useful about the song. I suppose a few decades ago, when anything electronic consisted of nothing more than FM beeps and bleeps, it may have had pertinence. Nowadays though, even modern rock songs, hip-hop, and soft new age songs make liberal use of synthesizers and post processing. So then, if all other genres are potentially a subset of electronic, what good is having it?
I really hope that Pandora's classification system clarifies this disorganized taxonomy we currently have (or, lack).
2008-12-06: Toward better temperature scales
Why do we still use silly temperature scales that allow such nonsense as negative readings? Both Celcius and Fahrenheit suffer this same defect, a non-absolute zero offset. Clearly (for anyone who has studied blackbody radiation, thermodynamics of gas expansion, or Brownian motion), an absolute temperature scale would ease the introduction of these concepts and simplify the calculations. So then, why don't we use one in every day use?
Let's look at the two best known candidates, Kelvin and Rankine. Since both are zero-based, they only (thankfully) differ in their scale factor, which affects both the values assigned to anchorpoints (like the freezing of water, a reasonable liquid to anchor to for any concerned organic life) and the range of common meteorlogical temperatures. The one reason why I can think that we don't use them in daily conversation is that both yield cumbersomely large numbers for common temperatures. For example, the freezing point of water is 273.15 for Kelvin and 491.67 for Rankine, while room temperature is 295 and 530 (though, Rankine does has an advantage that everyday temperatures are widely enough scaled to dispense with decimal points).
So, how about we pick a factor with (1) a large enough scale to make daily temperatures reasonable and decimal-less while still (2) anchoring it to something useful and easily remembered, like the triple point of water (close to the freezing point at sea-level atmospheric pressure, 0.01C). If anchored to exactly 1000 (1 kilodegree?), then most of the time you could abbreviate temperatures to the last 2-3 digits and yield a set of easy to remember temperatures like: 0-frozen water, 50-cool weather, 80-room temperature, 100-pleasant warm day. Now this would put the boiling point of water at 366 (1366T unabbreviated), but I'd claim that nice round numbers for boiling isn't nearly as key, since the boiling point any liquid is far more variable anyway - simply climbing a mountain will change it. In fact, climbing a mountain could reduce it to 360, which, for those who like nice numbers, is a fairly nice number. :-)
After some looking, I'm apparently not the first to think of this (or very similar). Alexander McAdie proposed that "the melting ice at a pressure of 1000 kilobars and is marked 1000" ("A New Thermometer Scale" by Alexander McAdie in Proceedings of the National Academy of Sciences, 1916-12-01, page 670-672, received 1916-10-23, from Blue Hill Observatory, Harvard University / related article: Physical Review Vol. VI, No. 6, December, 1915). Stephen Clif Brown also proposed (2003-09-12) that the triple point of water be anchored to 1000 (http://home.comcast.net/~igpl/Replace_the_kelvin.html). Converting from kelvins (triple point = 273.16) to T's would be T = K * 1000 / 273.16, or about K * 3.66 (hmm, which curiously resembles the boiling point above?).
2008-12-05: Celsius really superior to Fahrenheit?
First off, claiming that Celcius is actually part of the metric system is simply ridiculous, as Celcius is no more "metric" than old Fahrenheit. Every other respectable unit (meter, liter, gram...) in the otherwise clean system is an absolute scale - you don't start measuring weight from 100 kilograms (though I imagine some dieters might favor that).
While Celcius does have the initially seemingly nice notion that the range between water freezing and boiling is 100 units, in everyday, meteorlogical use (when someone asks how hot the weather is), Fahrenheit actually has a more useful real-world scale factor, where 0 means really cold and 100 means sweating hot. Celsius has a smaller scale factor, leaving it with lower precision for integers; and while technically, neither has any more/less precision than the other, Fahrenheit's finer granularity lets it reasonably dispense decimal values.
Celsius does have the winning point that it is simpler to do scientific calculations with, since conversion to a real scale, Kelvin, is just an addition. Though, that's arguably only because Kelvin was scaled to match Celcius, and Fahrenheit would have been the preferred one if Rankine had been adopted instead.
All that said, Fahrenheit starting at 32 degrees for freezing has always bothered me, and Fahrenheit certainly isn't superior to Celcius, but ultimately, both non-absolute scales are defective.
2007-11-17: Word Wrapping
Why is it called "Word Wrapping"? Sure, it makes some sense in English, but it's silly for languages like Japanese and Chinese, where line breaks can occur anywhere. It's silly for math equations, since it's not word wrapping, but variable and expression wrapping. It's silly whenever a line break occurs at punctuation. It's silly when the constraining text layout is smaller than the largest word, in which, it becomes character wrapping. Really, it's not an on/off thing, but more of an enum with multiple modes/flags (pay attention to explicit breaks, constrain to layout, never break words, optimize).
2007-01-01: XML treading where no markup should go
XML is not the answer to world's data interchange problem. It's just an answer, one of many. While it's very appropriate for text-centric languages like HTML (bulk is text, and the markup "mark's up" that text), it becomes overly verbose and illegible for more data-based formats, like configuration files (bulk is key/value pairs, with little document text). There is no such thing as one-size-fits-all, so one should pick the appropriate tool for the job. Hmm, what's that tool though?