Thursday, December 30, 2010


This is bugging the hell out of me. You log into this web service, right, and what greets you? A screen asking you to do something you don't want to do in the name of security, and this doesn't go away until you do what they want, resulting in a sort of extortion scenario where you have to give in to their dubious proposition in order to be left alone.

These two haunt my web experience:

  • Youtube wants me to link my gmail account with my youtube account for "security". There is no "Sod off, I don't want to."-button, just one that has it pestering me about it again the next time I log in.
  • Facebook wants to know my cell phone number and another email address. For "security". Every time I log in, I get this screen. There doesn't seem to be any way of getting rid of it.

No, I'm not going to give in to their pestering. It's just going to drive me away from their services.

Wednesday, December 8, 2010

Grand rant about non-physicists and their misconceptions and attitudes

These misconceptions and attitudes generally held by non-physicists have been bothering me for quite some while. So I'll list them here.


A peculiar error by people who do not understand physics is to attach "energy" to things for no reason. Don't get me wrong, if a machine does work, clearly some sort of energy transfer is going on, but this is implicitly true, so there's no need to say a magnet lifts things with magnetic energy. It lifts them with magnetic force1. It's also okay to just say magnetism.

There's also an alarming number of people whose idea of energy is alarmingly similar to old aether theories, as though it's some sort of incandescent medium that permeates reality.

The amazing pi!

"I'll memorize hundreds of digits of this mysterious, never ending number!" It's just a constant, and for almost all applications, you'll never need more than around 10 digits (say you make a meter-wide circle and want to know the circumference, then the error in your answer will be on atom scale if you have much more than 10 digits of pi, and at that level, measurement errors will dominate). Therefore, the irrational quality of pi is a curiosity at best. Nothing to write home about.

Furthermore, 2pi is the better constant. At almost no place in mathematics does pi appear where a constant equal to 2pi wouldn't make more sense.


Einstein did a lot of great things for certain, but his fame is way disproportionate to what he actually did. He's become some sort of patron saint/greek style personified concept deity of intelligence. People call you "an Einstein" when you're reasonably smart and/or a physicist. He did a lot of groundbreaking work, but this cult of personality is way out of hand (besides, everyone knows Feynman's cult of personality is the one to root for.)

If anyone deserves to be the patron saint of intelligence, perhaps C.F. Gauss is a better choice, seeing as how you can't swing a bracket in mathematics without hitting something named after him. Or Newton for that matter.

Schrödinger's cat

Erwin Schrödinger was one of the founding fathers of quantum mechanics. But what is he remembered for? His thought experiment about a cat. And this is a fairly questionable thought experiment that overall probably gives you more misconceptions than it does insights. I really wish people would think "equation" and not "cat" when they hear his name.

1 Someone is bound to sit at home with a copy of Griffiths' Electrodynamics in their hands all riled up about how it says the magnets do no work. This is an incomplete statement. While the Lorentz force does no work, spin is none the less acted upon by a force proportional to ∇B.

Tuesday, December 7, 2010

System instability, revisited

So, my system has been fairly stable since the last incident, but now it's taken another step on it's slow journey towards oblivion.

This time I was visited by seasonally appropriate graphical artefacts (i.e. small white boxes appearing and disappearing randomly). I was running the system under average to high load when they paid a visit. Switching from proprietary nvidia drivers to "nv"-drivers fixed it slightly, but it quickly turned worse so I rebooted. The artefacts mostly vanished, but some persist around alpha-blended graphics. Edit: This appears to be some sort of CPU saving mechanism in KDE 4. It mostly happened around widgets, but re-sizing them caused it to vanish.

Fig: Residual graphical artefacts. The entire screen looked like this, or worse. I tried to screenshot the mess at the time, but at that point, it was so bad I couldn't distinguish any dialogs, and therefore had no way of saving it.

This would indicate that the source of the problem is either the graphics card, or the memory corrupting the nvidia driver.

I'll run some more memtest86, and if that doesn't find anything, do a GPU transplant.

Saturday, December 4, 2010

Very peculiar system instability event

Just now I came home from some grocery shopping, popped the dinner into the microwave, and turned on my main desktop computer. From the LILO-menu, I decided to boot Windows. Got as far as "Disk Read Error", and then the system rebooted.

Ok, I thought, well, that drive is getting a bit old. But now LILO didn't even boot, I got the good old "L01 L01 L01 L01"-of-death screen. Not just a bad block, but some sort of progressive failure?

Okay, I thought, maybe I'll try the other LILO install I have on another MBR in case of emergencies like this. "L01 L01 L01"... and then it surprised me by actually loading, but that was as far as it went, as it behaved very erratically. I didn't manage to actually boot anything, and even navigating the menu was sluggish.

Now, the prospect of simultaneous disk failure seems a bit far fetched, and no other electronics have failed so I didn't consider a power spike particularly likely either. The CPU seems fine, as BIOS worked without trouble during the incident, and system temperature was also fine so a fan failure doesn't seem a likely explanation either.

Anyway, I'm digressing from the story. At this point I shut it down and let the computer rest for a while as I ate my now cooked dinner. Then I tried again, got as far as "L01 L01 L01" and then the main LILO suddenly worked. I managed to boot into Linux, and there were no obvious signs of error. So I rebooted into Windows, and it worked too.

I have no idea what this event was about, the most plausible explanation I've come up with is some sort of interference from the microwave (which I use on a daily basis, typically with the computer on and with no issues).

So fingers crossed that this was some sort of fluke event that won't repeat itself, but in case it was it's way of saying "Goodbye cruel world!", I've made several off-site backups of everything of importance.

Edit: Additional tests

I've performed some diagnostic tests on the machine, ran a bad block search with fsck and a memtest86 sweep. No errors. But still some residual instability. It seems that if I get past booting, everything is stable and okay, but there's a chance things go bad at the boot loader or during the kernel loading. Maybe the MRB, or some other early block (that wouldn't have been tested by fsck) is damaged?

Thursday, December 2, 2010

Base inversion and digit ordering

Here's a mathematical curiosity I stumbled onto. The direction in which we write our numbers is just a convention, and as is the numeric base we write them in. But the two are actually closely related. A switch in direction is equivalent to inverting the base.

Consider a number in base 10.

123.410 = 1 × 102 + 2 × 101 + 3 × 100 + 4 × 10-1

In base 1/10, this would be written as

4.3211/10 = 4 × 10-1 + 3 × 100 + 2 × 101 + 1 × 102

Clearly, these are one and the same.

That is all.