2/27/2010

02-27-10 - Good Products

I often rant about the shitty retarded modern product design that cruds things up with useless features while ruining the basic function with flimsy unpleasant construction. I though I should give a shout out to some things that get it right.

The most pleasing products in my life are simple, solid, well made analog old-fashioned tools. It's amazing the pleasure in doing something with a good tool; maybe the pleasure is just because I spent so much time with shitty products that when you use a good one it's like sweet relief.

Probably the happiest purchase I ever made in my life was a Japanese Chef's Knife . It just makes me smile every time I chop with it.

Recent good ones :

The Moroso Tire Pressure Gauge is a delightful bit of kit. Just rubber tubing and an analog dial, but it's such a massive improvement over those shitty stick-style gauges. Also, it's not just a gauge, it has a very fine air release valve, so you can get your tire pressures just right by slightly over-inflating them then using the Moroso to slowly release air until it hits the target number dead on. So satisfying.

( BTW while I'm sort of on the topic - I think I've said this many times before, but tires are by far the biggest bang for the buck in car performance ; it's just insane to care about your car's performance and not put very good tires on it. Furthermore, if you're going to do one fiddly thing for your car, that one thing should be adjusting tire pressures for load, temperature, usage ).

One of the insane classic examples is a good stapler . Fucking hell, a proper wonderful solid metal stapler that will last your whole lifetime costs $10-$20. And yet people scrimp in the stupidest fucking way and buy a $5 plastic shitty stapler that's a pain in the ass every time you use it, jamming and failing to puncture thick stacks of paper. This is a mistake I made plenty of times in my youth because I thought I was being thrifty, but it's just retarded, it's not thrifty, it winds up costing you more money in the long run (though only retards use that as an argument since money has different values at different time) and it costs you time and annoyance due to its shittiness. If you're buying basic things like a hammer or a wrench you should buy a really good basic one that will please you and last.

2/25/2010

02-25-10 - Espresso Girls

Seattle is now (in)famous for its bikini espresso stands. The five risque Everett Grab-N-Go baristas that were charged with prostitution are about to go on trial. The prostitution charged is based on showing their bits and performing some mild acts when the police tossed money at them, which apparently was common practice. Also common apparently was a game of "tip basketball" where girls would hold their bra or thong open to make a basket that patron would throw at from their car window.

We were in the national news for this before, back when Yakima banned thongs and see-through dresses for women ; it was a draconian across the board anti-thong law brought on by espresso stands. Several other area cities have passed laws limiting exposure and individual girls have been charged before for wearing only a thong and X-shaped pasties for example.

Apparently sexy espresso stands are a booming business and lord knows it isn't for the espresso ; Seattle is known for its coffee quality, but the swill that 99% of Seattleites drink is roughly on par with a Denny's coffee. Yes your downtown urban hipster foodie metrosexual will only buy beans that were grown in the shade under Juan Valdez' mustache, but the fat pasty turds in the suburbs just want some caffeine and sugar.

Of course the coffee girls are becoming minor internet celebrities and we can predict a coffee girls pay nudie site to pop up any day now. I'm sure all the news attention is only helping business; newspapers love these stories because it gives them an excuse to post girlie pictures.

I for one am totally in favor of the lascivious espresso stand. I have yet to actually see one myself; I guess they're all out in the far suburbs. Lord knows living in Puyallup or Everett is an excruciating punishment to all the sad sappy fucks who slog through life in the rainy styx - let them have a little T&A with their caffeine to brighten their day. It's a little odd that you're allowed to get high on chemical stimulants to your heart's content, but god forbid you look at a little flesh.

This is the fall of the roman empire! Live it up! I find it sad that for all our excess and indulgence and commercialism and connoisseurism , the pleasures that we are allowed to explore are so tightly bound and formulaic. Americans are absurdly rich, so much so that we basically make up trends in order to find ways to waste money, because we just can't come up with enough ways to waste money in the world. So we have created the whole "foodie" movement so that there is now plenty of $50 olive oil for you to waste money on, and we created the wine movement so you can waste money on $100 bottles - these are our pleasures, we summon the finest goods from all over the world back to Rome to satisfy us. But it's almost entirely limited to tastebuds. What about the sense of touch? What about sex, nudity? Our society is so puritanical and boring.

Furthermore, with the decline of the American education system and the laziness of wealth, what else are these girls supposed to do? They have no skills, there are no manual labor jobs for them to work in. The average American is destined for a life in a service industry, providing some sort of pleasure to the rich overlords. Let the young girls make money from their bodies before they get droopy and are forced into a life of foot massage or real estate sale makeovers.

02-25-10 - Laptop Death

My ancient lappy is trying to die. Much like my car that committed sepuku by being in the intersection when the Old Lady of Vengeance ran through, my laptop is trying to kill itself to force me to hurry and buy a new one like a normal person.

Roughly one year ago I looked at laptops before and concluded that they weren't enough of an upgrade over my belove AOpen 1557G from 2004 which has a lovely matte screen, the perfect laptop resolution at 15" of 1400 x 1050 , a good keyboard layout, and a Radeon 9600.

Now my laptop's VGA out is flaking out. It keeps dropping signal and I have to fiddle with the connector a bit. I think the problem is the solder points holding the VGA out to the board are loose. I could crack it open and try to fix them, but I feel like I have a 50% chance of completely breaking it if I do that (at the moment it still works perfectly if I use its screen as the display, but I almost never actually use the laptop as a laptop since laptops = body death. I use it as a portable desktop that I plug into one of my various workstations, and as an occasional emergency use on actual lap).

I'm terrified of going to a new machine. I have so many settings tweaked out on this thing that I could never hope to transfer. (god damn you Registry, god damn you).

I'm getting old, and computer revs are slowly killing me. Each time I move to a new OS or hardware platform, I spend less time getting to know it. I become more of a casual user. I can't be bothered to dig too much under the hood. Some aspect bugs me and I decide to just live with it rather than fix it. New CPUs come out and I don't learn all the nitty gritty details of how the pipeline works and how to write the fastest code for them. My expertise peaked about 5 years ago and is steadily going down hill, and I don't really see an end to it because honestly I just can't be bothered.

As an aside, I also used to be really hopeful and utopian about computers. I thought yeah the world out there is shit, but on computers it's a meritocracy and an amazing democracy, and researchers share information with hobbyists, and the internet is free for all, and it's only getting better and better. I think that trend peaked long ago and is now headed downhill fast. Yes, there is still a great community of people who are fighting hard to carve a utopian niche out of the electronic landscape, but they are more and more the minority, and things like Apple's draconian control of their platform, or the DMCA and HDMI content lock-downs, or putting advertising on your web site, or software you can't own but have to subscribe to - these are all things that would have gotten computer users outraged in 1995 and are just part of normal life now.

02-25-10 - Download MP3's from Youtube

Lately we're on a kick of listening to rare 70's hard rock and prog (which is fucking fantastic shit by the way). It's hard to find the recordings anywhere except Youtube, so I've gotten the need to turn the audio into MP3's.

So far as I know there is no good system to do this. (someone correct me if this is wrong). There are lots of service web sites that will do this for you, but they are slow and full of annoying adds and shit like that. The ideal thing would be a Firefox addon that just has a "save audio on this page as mp3".

(ADDENDUM : I guess there is one called "DownloadHelper", but they want money for the MP3 conversion feature. DownloadHelper does seem to be a little better than Unplug in that it has special code for Youtube that identifies the various format streams available so you can download the best one. Amusingly DownloadHelper seems to be very mainly targetted at porn site scraping; it has features like automatically finding new videos from certain sites and putting them in your queue for download. I guess it's the new era version of the automatic NNTP downloader I wrote in my youth. It looks like the FLV's generally have worse audio than the MP4's, so you should prefer them when possible perhaps.).

What I'm doing :

I use "Unplug" to download the .FLV video from Youtube. (* use DownloadHelper to download .MP4 video instead).

I use "MPlayer" to get the WAV audio out of the FLV. The way you do that is :


mplayer.exe -ni -vo null -vc dummy -ao pcm video.flv

Then I use the LAME command line build to encode the WAV to mp3. BTW there's a ton of bad old advice out there about LAME command line settings; I use this :

lame -h -V1 audiodump.wav r:\audiodump.mp3

but sometimes I wonder if I should use this :

lame -h -V1 --preset extreme audiodump.wav r:\audiodump.mp3

which actually makes smaller files usually because it enables some funny filters or something. Dunno.

Anyway, as a compression worker it bother me morally that I am decoding compressed audio from the FLV and recompressing it - I'd really like to just rip the audio stream directly from the FLV and be able to play that, but the fact is that my iPod and my car want MP3's, not AAC or whatever the fuck is in FLV, so it's pretty hopeless.

It's also annoying that I have to then enter the IDV tags by hand. I wish I could scrape the info right off Youtube. In theory you could use a CDDB auto-tagger on the MP3 to do it, but here are two problems with that : 1. I can't find a command line CDDB auto-tagger out there, there are several of them but they are all GUIs, and 2. even if you fire up the GUIs, CDDB seems to just spew fail all over itself.

ADDENDUM : if you don't need MP3's it is actually very easy to get the audio track directly to an M4A which can be played by iPods & most media players these days. Use mp4box ; I suspect there's a more direct command line that will work but this is the way that I've succeeded :


mp4box -new out.m4a -add gremlins.mp4#audio -ipod

2/24/2010

02-24-10 - Police Laser and Countermeasures

I did a bunch of research into police laser (lidar) and countermeasures, so I thought I'd share.

Police laser is infrared and is a pretty wide beam. At 1000-2000 feet (typical shooting distance) the beam will be a few feet wide. Unlike radar guns, they have to be stationary to use and can't shoot through glass, so the cop actually has to get out of the car and stand there and aim at you specifically. The laser has to bounce off something on your car and get back to the gun to be picked up.

Modern detectors like the V1 can detect laser, though on its own this is basically useless because when it alerts you have already been shot. Unlike radar, there is no detectable scatter. The V1 also false-alarms from sudden changes in illumination such as a reflection of the sun shining on it.

Cops are taught to aim at the license plate, but it's a myth that the plate is the only thing that can be picked up, it's just a good reflector and also a good center target (sort of like aiming your gun at the chest). Most cops and old guns only shoot cars from the front, but new guns and new technique will shoot the rears of cars. Many semi-intelligent people think that laser guns require a straight facing surface to get a reflection back, so they use tricks like angling their plate downward toward the ground. This does not work, because many surfaces on cars are retro-reflective ; that is, they have a BRDF with a large lobe directly backward towards the direction of incident light. License plates typically have retroreflective paint on them, and car headlights and tail lights are parabaloid mirrors which makes them great retroreflectors.

What can you do to avoid laser tickets?

1. Speed in the rain. Since cops actually have to get out and stand and shoot you, they don't do it in the rain. They would rather sit in the car and use the radar gun.

2. Stealth your car. This is theoretically possible to do, but very hard, and most of the commercial countermeasures are snake oil (things like the "Veil" diffusing paint coating or "Laser Shield" prizmatic license plate cover (Veil is basically sticky plastic goo that gives surfaces micro dimples so they scatter light randomly, Laser Shield is ridged plastic plate cover so it reflects light out to the sides)). To really successfully stealth your car you would have to do something like this :

Buy a car that's as small and curvy as possible (eg. an SUV is hopeless), you want lots of angled surfaces like a stealth fighter. Paint your car matte black (no gloss clear coat that will retroreflect). Replace your headlights with bright LEDs or something that doesn't use mirror reflectors. Replace all the glass with very flow reflectance or diffusing plexiglass. Remove your license plate, photo copy it and print it out on hard plastic or something so you eliminate the retroreflective paint.

While this is possible, it's pretty ridiculous and the halfway measures do almost nothing. Also some of those steps are of course illegal, but the thinking in the laser avoidance community is that it's better to get a non-moving-violation ticket for an illegal license plate or lights instead of a moving violation for speeding.

3. Install jammers. These do seem to be real. Radar jammers are illegal everywhere in the US, but laser jammers are legal in most states. Basically they work by having an IR emitter in the same band installed all over the car which makes the gun show an error code. The recommended technique is to use them with an on/off switch and a detector like the V1, so when you get a laser detection, you slow down then turn off your jammer so that the cop can then read you at legal speed.

Laser jammers do not work if they are mounted concealed, they must be mounted with the heads prominent, preferrably 4-6 heads at key points on the front and rear. The best jammer appears to be "Laser Interceptor" , and that runs about $2000 installed.

Cost/benefit says the jammer is not worth it unless you already have several tickets and are at risk of going into the hazardous driver insurance pool or losing your license. (or live in a state like Oregon where it's hard to use a lawyer to get out of tickets).

BTW as an aside it annoys me that worrying about cops and fiddling with my radar actually makes me a much more dangerous driver than I need to be. I try very hard to eliminate all distractions and just focus on the road and the other cars and the driving, but such a huge amount of my attention has to be spent keeping my eye out for cops or responding to radar blips. It's also a myth that tickets make people safer ; boy racers don't slow down until they have enough tickets to lose their license. I never slowed down when I was young and paying $6000/year in insurance. The penalty is just not high enough, and not predictable enough.

I'm a much safer driver now, but the reason is crashing a number of times. None of the crashes was due to me speeding, they were largely due to things out of my control, such as bad road condition problems or old ladies running red lights. The thing that really did it though was my cycling crash, which made me realize how absolutely fucking shitty it is to get injured and have to live the rest of your life with a damaged body.

2/23/2010

02-23-10 - Image Compresson - Color , ScieLab

The last time I wrote about anything technical it was to comment on image coding perceptual targets and chroma . Let's get into that a bit more.

There are these standard weapons available to us : 1. Colorspace transform (lossy or lossless) , 2. Relative scaling of color channels, 3. Downsampling , 4. Non-flat quantization matrices.

Many image compressors use some combination of these. For example, JPEG uses YCbCr colorspace, which has a built-in down scaling of the chroma channels, also optionally downsamples chroma, and also usually uses a very high-frequency-killing quantization matrix. The result is that chroma is attacked in many ways - the DC accuracy is destroyed by the scaling in the color conversion as well as the [0] entry of the quantization matrix, and high frequency info is killed both by downsampling and the high entries in the quantization matrix.

But is this good? Obviously it's all bad in terms of RMSE (* not completely true, but close enough), so we need something that approximates the human eye's less sensitie chroma receptors.

For a long time I put off this question because it seemed the only way to attack it was by showing a ton of images to test subjects and asking "is this better?". (Furthermore, there's the ugly problem that any perceptual metric is heavy tied to viewing conditions, and without knowing the viewing conditions you may be optimizing for the wrong thing). But maybe I found a solution.

Let me be clear briefly that I am here only trying to address the issue of how the human eye sees chroma vs luma. This is not a full "psychovisual perceptual metric" which would have to account for the brain identifying areas of noise vs. areas of smoothness, repeated patterns, linear ramps, etc. Basically the only thing I'm trying to capture here is the importance of luma bits vs. chroma bits.

Well, it turns out there's this thing from color research called SCIELAB . You may be familiar with "CIE LAB" aka the "Lab color space" which is considered to be pretty close to "perceptually uniform" , that is 1 unit of distance between two Lab colors has the same perceptual error importance no matter what the two colors are. Well SCIELAB is the extension of CIELAB to images (not just single colors). You can read the paper at that link (or see links below), but the basic thing it does is very simple :

SCIELAB takes the image and transforms it to "opponent color" (luma, red-green, and blue-yellow) , which is roughly the color space that eyes use to see light (rods see luma, cones see chroma) (note that here we are transforming "pixel values" into real light values, so we have to make an assumption about the brightness and color calibration of the viewing device). In opponent color space, each channel is filtered. The filter used for each channel represents the angular resolution that a rod or cone has. Basically this is a gaussian whose sdev is proportional to the angular resolution in that channel. This depends on the DPI of the viewing device and the viewing distance (eg. how many pixels fit into one degree at the eye). The gaussian is narrow for luma, indicating good precision, and wider for chroma. The filter also has a wide negative lobe around the center peak, which captures the fact that we see values as relative to their neighborhood - eg. 100 on a background of 10 looks brighter than 100 on a background of 50.

The gaussian filters represent the probability of a photon from a given pixel hitting and activating a rod or cone. The wider filters for chroma indicate that a half-toned image in red-green or blue-yellow will be indistiguishable from the original at a much shorter distance than a half-toned luma image.

One you do this filtering, you transform back to CIELAB and then you can just do a normal MSE to create a "delta E". (CIE also defines a more elaborate more uniform "delta E" metric for LAB , but for our purposes the plain L2 distance is very close and much simpler). The result is a "SCIELAB delta E" metric that is analytic and can be used in place of MSE for comparing images. Having this SCIELAB metric now lets us try various things and it tells us whether they are perceptually better or not (in terms of optical perception of color, anyway).

So far as I know this has never been used in the mainstream image compression literature ; the only place I found it was this Stanford school project tech report : Direction-Adaptive Partitioned Block Transform for Color Image Coding . This paper is pretty interesting; they aren't actually doing anything with the DA-PBT , they're just evaluating color spaces and how to do color coding starting with a grayscale image compressor.

Let's go through the EE398 paper in detail.

First they use YCbCr because they claim it produces better scielab results than RGB. True enough, but there were a lot of other color spaces to try. Furthermore, they don't mention this, but they are using the JPEG style YCbCr, which has a built in 0.5 scaling of the chroma channels (chroma should have a range of [-256,256] but JPEG offsets and scales to put it back into [0,256]) - they have effectively killed the chroma precision by using YCBCr.

They then look at whether sub-sampling helps or not. They find it to be roughly neutral - but when you try subsampling or not subsampling you should also try optimizing all other free options (scaling of the chroma channels, quantization matrix).

The most interesting part to me is "Rate Allocation". They try giving different fractions of the bit budget to Y or CbCr. They find that optimal delta E almost always occurs somewhere around Y bits = 66% of the total , that is the bit ratios are like [4:1:1]. In order to acheive this ratio they had to use small quantization step sizes for CbCr than Y, but that is an anomaly because of the fact that the YCbCr they use has killed the chroma - if you use a non-scaling YCbCr you would find that the chroma quantization values should be *larger* than luma to acheive the 66% bit allocation. (note that using different quantization values on each channel is equivalent to scaling the channels relative to each other).

They also found that using non-uniform quantization matrices (ala JPEG) hurt. I believe this was just an anomaly of their flawed testing methodology.

This paper was the most serious study of color in image compression that I've ever seen, but is still flawed in some simple ways that we can fix. The big problem is that they make the classic blunder of many people working in compression of optimizing parameters one by one. That is, say you have a compressor with options {A,B,C}. The blunderer finds the optimal value for option A and holds that fixed, then the optimal for B, then the optimal for C. They then try out some experimental new mode for step A, and their tests show it doesn't help - but they failed to retry every option for B and C in the new mode for A. eg. for example something like downsampling might hurt if you're using YCbCr, but say you use some other color space, or scale your colors in some way, or whatever, then downsampling might help and the result of doing all those steps together may be the best configuration.

Let's go back through it carefully :

First of all, the color conversion. Let me note that we use the color conversion in image compression for really two separate purposes which are mixed up. One use is for decorrelation (or energy compaction if you prefer) - this helps compression even for lossless mode. The second is for perceptual separation of chroma from luma so that we can smack the chroma around. Obviously here we need a color transform which gives us {luma/chroma} separation - that is, we cannot use something like the KLT which doesn't necessarilly have a perceptual "luma" axis.

From my earlier color studies, I found that YCoCg produces good results, usually within 1% of the best color transform on each image, so we'll just use that. But we will be careful and use a float <-> float YCoCg which doesn't scale any of the channels.

We will then scale Y relative to CoCg. This scaling is equivalent to variable quantizers and is (one of the ways) how we will control the bit allocation to Y vs. Chroma. This scaling gives you a difference in "value resolution" , it doesn't kill high frequencies.

You can then optionally downsample chroma. Note that in naive tests I have found in the past that downsampling chroma sometimes helps visual quality; and in fact in some cases it even helps MSE measured on the RGB data. I now know that that was just an anomaly due to the fact that I wasn't considering chroma scaling. That is, downsampling was just a crude way of allocating fewer bits to chroma, which does in fact sometimes help, but if you also have the ability to change the chroma bit allocation by relative scaling of the channels, the advantage of downsampling vanishes.

I optimized the scaling of CoCg relative to Y on lots of images. Obviously the true optimum value is highly image dependent (you could compute this per image and store it with the image of course), but in most cases a scale near 0.7 is optimal if you are not downsampling, and a scale near 1.1 is close to optimal when downsampling ( 1.0 is not bad when downsampling ). When not downsampling, the optimal bit allocation is usually in the area of Y ~= 66% of the bits, as seen in the EE398 paper. When downsampling, the optimal bit allocation tends to be closer to Y = 80% of the bits. Downsampling generally hurts RGB MSE and SCIELAB delta E, but I find it sometimes helps RGB SSIM.

Obviously downsampling is resulting in more bits being used on luma, which means you'll have sharper edges and better preservation of texture and a visual appearance of more "detail", at the cost of the color values being far off. By my own examination, I often will find that if I just stare at the image made from downsampled chroma it looks "better" - eg. I see more edge detail, and it has less of that obvious appearance of being compressed, eg. less ringing artifacts, halos, stair-steps, etc. However, when I switch back and forth between the original and the compressed, the version made from downsampled chroma shows obvious color errors. The version made from non-downsampled chroma obviously has much better color preservation, but appears generally blurrier, has more block artifacts, etc. The non-downsampled version wins according to "delta E" , but by my eyes I can't really clearly say one is better than the other, they're just different errors.

The last tool we have is a non-uniform quantization matrix. NUQM lets us give more bits to the low frequencies vs. the high frequencies. Generally NUQM hurts MSE, but it might help "delta E" , because SCIELAB accounts for the "fuzziness" of human visual (insensitivity to high frequency pattern). To test this, what we need to try is various different NUQM's for both luma and chroma, as well as optimizing the relative scaling value in each case. I haven't completed this yet, but early results show that NUQM's do in fact help delta E. Note that I'm not talking about doing a per-image optimal NUQM like "dctopt" does or something, just finding something like the JPEG style skewed matrix to use globally.

Some numbers for example :


On a 512x512 color image of a face , at 1.0 bits per pixel , 
optimizing quality at constant bit rate


baseline : delta E = 2.2933

not downsampled , optimal CoCg scale = 0.625 : delta E = 2.155  (bits Y = 72%)

    downsampled , optimal CoCg scale = 1.188 : delta E = 2.381  (bits Y = 80%)

best NUQM and scaling (no downsampling) : delta E = 1.899  (bits Y = 61%)

( JPEG delta E = 2.7339 )

One thing I notice that NUQM does obviously is give a lot more bits to the DC's. In this case :


not downsampled, same cases as previous
UQM = uniform quantization matrix

 UQM , bits DC = 15.4% , Q = 14.0  , delta E = 2.155 , bits Y = 72%

NUQM , bits DC = 19.4% , Q =  7.25 , delta E = 1.899 , bits Y = 61%

Here Q is the quantizer of the DC component of Y - in the UQM case all Q's are the same (though the Q for chroma is effectively scaled). In the NUQM case the higher frequency AC components get much higher Q's. We can see from the above that because of NUQM, the quantizer for the DC can be much lower at the same bit rate.

Personal visual inspection indicates that the NUQM images just have much more "JPEG-like" artifacts. That is, they generally look more speckly. They obviously preserve flat areas and simple ramps somewhat better. The tradeoff is much worse ringing artifacts and destruction of high frequency detail like fine edges. (in my case the lower Q from NUQM also means a much weaker deblocking filter is used which may be part of the reason for more speckly appearance).

In any case, NUQM clearly helps delta E due to the ability to take bits away from the high frequency chroma data - much better than just scaling and downsampling can.

This is all very interesting and promising, but we have to ask ourselves at some point - how much do we trust this "scielab delta E" ? eg. by optimizing for this metric are we actually making better results? More and more I am convinced that the biggest thing missing from data compression is a better image quality metric (and then once you have that, you need to go back to basics and re-test all your assumptions against it in the correct way).

Color links :

Working Space Comparison sRGB vs. Adobe RGB 1998
Welcome to IEEE Xplore 2.0 Using SCIELAB for image and video quality evaluation
Video compression's quantum leap - 12112003 - EDN
Useful Color Equations
Useful Color Data
Standard illuminant - Wikipedia, the free encyclopedia
SpringerLink - Book Chapter
S-CIELAB Matlab implementation
References related to S-CIELAB
Lab color space - Wikipedia, the free encyclopedia
IEEE Xplore - Login
help - sRGB versus Adobe RGB (1998)
efg's Chromaticity Diagrams Lab Report
CIECAM02 - Wikipedia, the free encyclopedia
Chromatic Adaptation
Brian A. Wandell -- Reference Page
Ask a Color Scientist!
A top down description of S-CIELAB and CIEDE2000. Garrett M. Johnson. 2003; Color Research & Application - Wiley InterScienc
A proposal for the modification of s-CIELAB

2/22/2010

02-22-10 - New Car

For those that don't see me in real life, FYI I got a new (to me) car :

P1010196.JPG.bmp

We flew down to Arizona to get it and drove it back to WA. Spent the first night in Vegas, then drove back the 93 to the 84. The 93 is perhaps the most desolate major highway in the US; it's all high desert, big flat valleys and stark snow capped mountains. There were a handful of fun twisty mountain passes, and some big straights where I got the car up to 130 or so. It was an excellent way to get acquainted with her. You can see an annotated slideshow at flickr .

(while I was at it, I also finally put up some Thailand pictures ; these are pretty much all N's nice photographs).

Of course I went through most of the US without much trouble from the fuzz, and immediately got a ticket in Oregon for going 85 in a 65. Fucking Oregon. I now have also found that Oregon is one of the worst places in the US to get a ticket, because they don't have a system where you can use lawyers to get off; apparently they have laws or precedents or whatever which means you can't do the normal thing that lawyers do of challenging the radar gun calibration. (of course, hypocrites that humans are, it's not like Oregon is just generally a more law-abiding place or a place where you can't get out of charges; try searching for "Oregon Lawyer" and you will find that the state is absolutely chock-a-block with drunk drivers who get out of their convictions, there are no speeding ticket lawyers, but masses of DUI Diversion lawyers). (see more at end)

How's the car? (you ask) , eh it's fucking great. Is it worth it financially? Surely not, you could be almost as happy in a 370Z or an Rx8 ; really the modern Japanese sports cars are such amazing value that buying anything else is pretty retarded. But I've never felt anything like it. The controls just feel so good, the cluth is so heavy and solid, the steering feedback is sublime, the gas pedal is responsive. The suspension is too stiff for the terrible Seattle road conditions, and it's not an ideal commuter car, but when a gap in traffic opens up and I slam the trottle, or when I go ripping around a corner, it makes me smile and everything is right with the world.

I've been joy riding around Seattle a bit, looking for fun roads. We went and cruised out Redmond - Snoqualmie Valley - High Bridge Road - Snohomish this weekend, which is one of the classic motorcycle rides in the area; not a bad run. On the way home for AZ, we took the 14 through the Columbia Gorge a bit, which was pretty sweet. There are some really classic roads in that area that I'll go back to some day ( Rowena Loops on Hwy 30 in Oregon, Maryhill Loop Road off the 14, and then you've got Wind River Road and lots of good stuff in the Gifford Pinchot, but that has to wait for summer). When you're looking for good driving roads, the best way is to look at motorcycle web sites, since those guys actually pay attention to scenery and pavement quality and twisties and all that good stuff.

More ticket related ranting : when I got the ticket in Oregon, I was of course travelling along with a few other cars at exactly the same speed. The cop singled me out and popped me with laser and came over and picked me out of the group. That's a fucking awful disadvantage of this car. There's a story going around the net that Geico gives millions of dollars worth of laser & radar guns to police departments. Dunno how true that is, but it is pretty sick the way the police are basically giving money to insurance companies. I actually don't have a big gripe with police using traffic stops to raise money for their cities. It's not their fault, it's the fault of all of you - the fucking retarded voters who vote against taxes and yet want services and give them no other ways to raise money. The sick thing is that even on a $300 ticket, after the cost of the police time and the court and everything, they probably only get $20-$50 for the city. On that same ticket the insurance company gets $2000 or so over the next five years.

Oh, also while I'm ranting, the fact that tickets are based on amount over the limit is fucking retarded. It should be more like *percent* over the limit. 85 in a 65 on a fucking wide open freeway is a big difference from 45 in a 25 , but both are tickets for twenty over.

Maybe I'll write up a full review of the car for the internet's benefit some day.

02-22-10 - Light Rail Racing

It's hard to find a fun drive in Seattle proper because the pavement is generally rubbish, it's too trafficky, full of really dumb pedestrians, and out in the East Side where the roads are better it's swarming with bored cops. But I found one fun game to play : Light Rail Racing.

MLK in south Seattle was recently all torn up and rebuilt; it has fresh smooth pavement, and the light rail line running down the middle of it. Late at night (which in Seattle means after 9 PM), it's almost completely empty of traffic, but the light rail is still running. When the light rail runs, it triggers all the lights ahead of it all down the street so that it doesn't have to stop. The light rail trigger puts up a flashing light and a bell and all kinds of stuff.

What you do is hang out on MLK until a train comes by, then you pull out and just surf the wave of greens all the way down the road. It's almost like a runway - long smooth empty expanse with all the cross roads blocked off.

02-22-10 - Computers in Cars

I really think computer screens in cars should be illegal. I know I'm a bit of a luddite and many of you are fans of this whole car gadgetry trend, but let me try to convince you of its stupidity.

For one thing, it's absolutely dangerous. There are tons of buttons, and pretty much all the car computers use some kind of selection wheel and context-sensitive buttons. That means you have to look at the screen to know what mode it's in to know where you should scroll and what the button will do. Nothing in a car should ever require you to take your eyes off the road. I know you get to memorize the menus a bit over time, but every car I saw the menus are very deep and complex, and people are inevitable seduced to go fiddle into their car status and check their average speed and tire pressures or go edit their address book during traffic, and that means unsafe driving. I think most people now finally would admit that talking on a cell phone in a car (even hands free) is very dangerous (though it doesn't seem to stop anyone from doing it) - the car computer is really no better.

It's a terrible trend that so many cars now have the radio and air conditioning and such settings buried in these computers, which means just for normal adjustments that you want to make while driving you have to look away from the road and fiddle around.

But say you just don't give a fuck about hurting other people or crashing yourself. I still think car computers are retarded. One problem is just that they are generally rubbish. Okay, let's pretend that's not a problem and your car actually has a really good one. You're still faced with what I believe is the most fundamental problem :

Car time scales are not the same as electronics time scales.

That is, in 5 years your car might still be very usable and great, but in 5 years your computer will be shit. Computers should not be tied to the car. In 5 years your nav system will be laughably poor compared to cutting edge, the display will suck, etc. etc. Even if you like car computers, they should be a separate pluggable component, not built in. Furthermore, car development times are generally long, so by the time the car comes out, the computers in it are generally already out of date.

Hell cars can be kept 10-20 years, and in those cases the computer becomes a really annoying anachronism.

2/16/2010

02-16-10 - Linkage

YouTube - Air - La Femme D'Argent
ignore the song, it's a great video of Market Street in San Francisco 1905 , watching the traffic weave around

The Curry House - Glossary of Indian Restaurant Curries
pretty good page though probably the UK versions of names

Seattle Traffic stop Map
nice map, surprisingly gathered by the Post Intelligencer. You can clearly see the speed trap at 85th on the 405 right where our work is. It seems there are basically no cops on the bridges, so I can go ahead and speed there more.

Quick question for a Harrier driver, or JF [Archive] - PPRuNe Forums
Holy Shit - I love it when I find these little web forums that are tiny enclaves where the actual fucking people who do these things chat with each other. In this case, the thread is about the hover capabilities and limitations of the Harrier, but you can browse around the forum and be endlessly entertained.

Photo Shoots 2009
Nevada Northern winter freight weekend 1 photos
Nevada Northern 2007 Winter Freight
- another similar thing - there's this whole community of people who restore old trains, and then pose them and take photo shoots of them; it's a lot like pornography, the trains are just preened and posed to capture the "money shot". Very interesting.

Manheim Consulting Used Vehicle Historical Index
Unforunately they charge for their best data.

Kevin Beason - gobal illumination
Lots of papers here.

HEAVY LITTLE OBJECTS
is a blog where a dude takes specimen-style photos of random objects around the world. The newer stuff is kind of bollocks, but if you browse into the past a bit there's some gold.

and finally, I heard Bruce Dixon on the radio and it made me think "amen brotha, testify!" ; he's the only speaker I've heard in a long time that really got me fired up and ready to take to the streets. One of the main things he's saying these days is basically that Obama is a huge disappointment to liberals and blacks in particular, and we need to hold him accountable. The Obama administration and policies are a direct continuation of GW Bush (same people running the Fed, same people running Defense) , and in some ways things are actually worse (civil liberties, transparency, executive immunity, illegal war methods). He doesn't just call out Obama though, his greatest ire is at the weak liberal voices that refuse to critize Obama, that are beholden to corporate interests, that fail to do independent journalism or speak the truth to power. Sample some Bruce Dixon here : Black Agenda Report the journal of African American political thought and action
Black Agenda Report Obama's SOTU, more than a penny short evildoer's Blog

02-16-10 - Chief Sealth

Who the fuck is Chief Sealth? It's popping up all over recently, with Chief Sealth Bike Trail and Chief Sealth High School. Apparently it's just the new popular anglicization of Seattle (which Wikipedia likes to call Si'ahl). Stupid, once you pick a screwed up anglicization, just stick with the same one. Ten years from now they'll decide Sealth isn't good enough and we'll get some new weird name.

Anyway, it got me to thinking - who really was this Seattle character that we're named after?

Seattle is much liked by hippies because of his famous speech which is full of wisdom and environmentalism and shit like "the land does not belong to us, we belong to the land". Apparently this speech are similar quotes are entirely apocryphal.

If you read between the lines of the official history a little bit, I think we can see who Seattle really was : an Uncle Tom, a conciliator, and a traitor. He was chief of the Duwamish, which was one of the weaker tribes in the area (weaker than the Snohomish or Nisqually). The various tribes often fought, and Seattle was wise enough to see that the way to strengthen his tribe was to do a deal with the devil and make friends with the white settlers who were moving in to the area in increasing numbers. So Seattle befriended Maynard and they used each other as allies to increase their personal powers. In the brief "Battle of Seattle", Chief Seattle held out his tribes and the other natives were easily defeated.

A better hero is Leschi , chief of the Nisqually, who was initially accomodating to settlers, but when it became clear that his tribe would be forced onto unlivable land, he tried to foment a rebellion from all the tribes. He was captured and killed by the whites of course.

Of course the whites wrote the history, and with very little first hand information or accounts from natives, Seattle was made the hero and the prominent native leader. So the next time you see a statue of Chief Seattle (or Sealth or Si'ahl or Seaillhth) , spit on it.

2/10/2010

02-10-10 - Some little image notes

1. Code stream structure implies a perceptual model. Often we'll say that uniform quantization is optimal for RMSE but is not optimal for perceptual quality. We think of JPEG-style quantization matrices that crush high frequencies as being better for human-visual perceptual quality. I want to note and remind myself that actually just the coding structure actually targets perceptual quality even if you are using uniform quantizers. (obviously there are gross ways this is true such as if you subsample chroma but I'm not talking about that).

1.A. One way is just with coding order. In something like a DCT with zig-zag scan, we are assuming there will be more zeros in the high frequency. Then when you use something like an RLE coder or End of Block codes, or even just a context coder that will correlate zeros to zeros, the result is that you will want to crush values in the high frequencies when you do RDO or TQ (rate distortion optimization and trellis quantization). This is sort of subtle and important; RDO and TQ will pretty much always kill high frequency detail, not because you told it anything about the HVS or any weighting, but just because that is where it can get the most rate back for a given distortion gain - and this is just because of the way the code structure is organized (in concert with the statistics of the data). The same thing happens with wavelet coders and something like a zerotree - the coding structure is not only capturing correlation, it's also implying that we think high frequencies are less important and thus where you should crush things. These are perceptual coders.

1.B. Any coder that makes decisions using a distortion metric (such as any lagrange RD based coder) is making perceptual decisions according to that distortion metric. Even if the sub-modes are not overtly "perceptual" if the decision is based on some distortion other than MSE you can have a very perceptual coder.

2. Chroma. It's widely just assumed that "chroma is less important" and that "subsampling is a good way to capture this". I think that those contentions are a bit off. What is true, is that subsampling chroma is *okay* on *most* images, and it gives you a nice speedup and sometimes a memory use reduction (half as many samples to code). But if you don't care about speed or memory use, it's not at all clear that you should be subsampling chroma for human visual perceptual gain.

It is true that we see high frequencies of chroma worse than we see high frequencies of luma. But we are still pretty good at locating a hard edge, for example. What is true is that a half-tone printed image in red or blue will appear similar to the original at a closer distance than one in green.

One funny thing with JPEG for example is that the quantization matrices are already smacking the fuck out of the high frequencies, and then they do it even harder for chroma. It's also worth noting that there are two major ways you can address the importance of chroma : one is by killing high frequencies in some way (quantization matrices or subsampling) - the other is how fine the DC value of the chroma should be; eg. how should the chroma planes be scaled vs. the luma plane (this is equivalent to asking - should the quantizers be the same?).

2/08/2010

02-08-10 - The Early Shift

I've been falling off the early shift wagon a little bit. I just can't figure out how to fucking LIVE !? In theory the early shift is pretty sweet. You get home by 3:30 or 4:00 so you have time to run around and exercise in the sun before dinner. The office is nice and quiet when you get in so you can focus.

The biggest win of the early shift is that traffic is so *reliably* non-existant. People often suggest shifting late as if that's equivalent to shifting early. It's not even close. The thing is, traffic problems are not of instantaneous duration. An accident during rush hour (or a bridge opening or whatever) will cause lingering traffic for *hours* afterward. That means when you shift late, you either have to shift *really* late, or you have to have a pretty high probability of at least moderate traffic.

The traffic density ramps up very fast in the morning, reaches peak in 7-8 and then bleeds off very slowly. (The best data I've found is UW HOV study ; click the green boxes to download PDF's or go straight to Congestion Frequency ). Congestion before 7 AM is reliably near zero. Then it trails off gradually.

To shift late to really avoid traffic you have to shift to 11:00 - 7:00 , and even then you're going to be hitting way way more traffic than you would at 7:00 - 3:00 ; something like an accident on the 520 can easily make traffic bleed way past 7:00.

So, shifting early is awesome. No, not really. It means you have to go to sleep by 10 or so, eat dinner at 6-7, so you never get to see friends at night. If you do stay up late one time, it completely fucks you up.

The worst thing is that my body clock just doesn't want to adjust even with about 2 weeks of pretty steady early schedule. I wake up and I'm just exhausted, I'm yawning all day. I don't feel hungry at the early-shifted meal times, but if I don't eat I get woozy. So I'm just a bit miserable all the time.

I don't know how to live!!

02-08-10 - Mike McGinn

I just heard a long interview on the radio with our mayor Mike McGinn (done by the hated Steve Scher). Wow, I'm impressed. He was very smart, direct, he listened to callers and turned their retarded emotional questions into rational interesting answers. A few things he said that I think were really cool :

He said Seattle is in a huge budget crisis. Obviously we have to find new revenue sources and also cut services. He generally didn't try to dodge or candy-coat the issue or make any stupid pledges to not cut this or not cut that the way most politicians do. One thing he said that was particularly good was that we need to find a long term sustainable budget balance where intake equals expenditure, as opposed to continuing the one-off tricks to make the budget work this year without thinking about the future (our previous government has done things like freeze hiring and freeze purchases of new buses - these things only push maintenance cost into the future).

He talked a bit about the whole viaduct broohaha. I actually didn't know that a big part of the problems it that the Seattle seawall has been deemed unsafe; engineers say it has a 10% chance of failure in the next 10 years, and in a big storm we could have a huge disaster like situation in the waterfront low areas. McGinn says the top priority is the replacement of the seawall whether we do a tunnel or not, so I think that's cool. He also said his vision for a non-tunnel alaskan way is a lot like what SF did with the embarcadero - just turn it into a regular street, and use the savings for public transit and improving I-5 as well.

They talked about the 520 bridge project a bit also. McGinn wants to have no HOV lanes on the new bridge and instead have light rail. Whether you agree with that or not (I do), his reasoning is just very mature and very sensible. 1. The reality is that HOV is not used as much as it needs to be to make sense as a dedicated lane, and buses can share the light rail lane. 2. we need to provide a means of transit for the poor in our increasingly 3rd-world country, and 3. this bridge will have to last 75 years, so let's think about the city we are building for the long term future.

However, before we get too excited, let us remember that McGinn has very very little power, the Seattle mayor is a weak executive; all the decisions will be made by the city council, and most of them have to be ratified by the state legislature and possibly the king county council as well. It's a huge beaurocratic mess of cross-organizational committees.

In these situations I wish we could just vote McGinn to be temporary dictator and let him fix our fucking mess without all the retarded interference from petty lawmakers protecting their pet interests.

ADDENDUM : I forgot the main reason I wanted to write about this, was that it was somewhat discouraging to hear the reasonable intelligent voice of McGinn contrasted with the self-serving retarded petty emotional callers. The typical caller had questions (all call-in questions are actually comments, of course) like "I live in Federal Way, why should I be paying for the viaduct?". Well, first of all Mr. Federal Way Retard, most people who drive on the viaduct either live or work outside of Seattle proper. It's not for people who live in the city to get around the city, it's a connector that goes right through and largely benefits the surrounding communities. In fact, all surrounding communities generally steal value from the big cities; the city centers are what provide jobs and population magnets, but they get less revenue per capita as the rich people move to the outlying suburbs, so if we were being fair the outlying suburbs should really be paying a lot more towards inner city projects. Second of all, even if it weren't a fair distribution of charges, can't you see the bigger picture? These are long term projects for the whole metro puget sound area; these are about constructing a city for all of our future. We shouldn't bogged down in these petty squables about how it will affect one particular house that's near the 520, we should be thinking about the greater good for the whole area 50 years down the road.

02-08-10 - Happy Jobs

The standard trite advice for youngsters considering various jobs is "do what you love" and "follow your passion" and "forget the naysayers and the people who want you to be practical".

What a load of poppycock.

The reality is that pursuing what you love is of marginal value at best, and is harmful at worst. Many people like myself find that going to work in something you love (video games) strips all the joy out of that thing and in fact removes something positive from your life. Usually doing "what you love" means going into a crowded field where lots of other people also want to work, which means you will have trouble getting a decent job and decent pay.

What actually matters to happiness in the workplace ?

1. Choose a job you can be very good at. There's just massive benefit to being one of the best in the world at something, or having a talent that's desirable or unusual. You don't want to be an easily replacable worker, because then you can be treated like this. This does not mean you have to be some superman, however. You could specialize in something that most other people don't want to do. Or you can get some licensing that makes you rare (like a crane operator or underwater welder).

2. Choose a job that doesn't tie you to a specific company or geographical location. Avoid jobs where there's basically just one employer in that field, or where all the companies are in the same city. Your power in controlling your life comes from the ability to move where you want to live, or change jobs if your employer treats you badly. You never know when something will go sour with a certain employer or you may have to or want to move for some reason.

3. Choose a job where the working conditions are pleasant. This means the physical environment of the office, the hours you have to keep, who your coworkers are. All of this is way more important than what you actually do for the work. A huge factor in this is the aforementioned freedom to choose from multiple employers - you don't want to be stuck having to accept the one job you can get, freedom to choose lets you pick an enjoyable place to be.

4. Choose a job with flexibility. This may vary for others, but for me the flexibility to work what hours I want, take vacation when I want, etc. is massive. Generally this means picking a job where you are judged by delivering some work product, not just putting in hours each day. Then you can choose your own path to make that work product. Obviously this could be software, but the same goes for custom bicycle frame builders, web site designers, etc. anyone who's running their own business or an independent contractor or just in a good employer situation where they give you product deadlines instead of work hours.

5. Choose a job with a pathway for advancement or a way to stand out.

6. etc... you get the idea

The important thing is freedom and power, not doing something you're passionate about. Passion is for love making and cattle. This is work.

2/05/2010

02-05-10 - Friday Linkage

Waterlines Overview - page about the water/landscape history of Seattle ; kind of janky flash interface but some good stuff

The Singer 911 - okay I don't actually think the 911 is a great example of this, but I stumbled on this and it reminded me that if I had arbitrary money and multiple cars, what I really want is an old car that's hollowed out and filled with modern bits. Old cars just look *SO* much cooler than new cars, I love old-style paint jobs, and they have better functional designs too, they're smaller and lighter. I *LOVE* the minimal functional interiors with just simple switches and no computer bits. And with a modern engine and the light old car, this thing does 0-60 in 3.9 sec. Fuck yeah. (instead of the 911 I'd rather see an old Ferrari, like a 1970 246 or Ferrari 250 GT SWB , or an old Jag like an XKE or XKSS, or a BMW 507, or there's plenty of gorgeous old cars to choose from )

The Empire That Was Russia The Prokudin-Gorskii Photographic Record Recreated (A Library of Congress Exhibition) - some cool photos or the diversity that was Russia before communism

Southwest Image Gallery - more cool photos; holy shit Southern Utah has an embarassment of riches, and despite my constant road trips around that area I have never been through the good stuff at all because it's always too far out of my way as I'm trying to do the CA-TX route. I would love to cruise all through there some day and do Bryce / Canyonlands / etc

Security Driver Article Info Center - holy shit holy shit. This page is written for like Blackwater driver type people. Some creepy shit in here, and also some really good articles on just how to drive well.

Pashnit California Motorcycle Roads - really nice motorcycle touring page. Has tons of photos and route slips. Useful for cars looking for nice country roads, and even for bicyclists. Mostly California. I miss CA.

English Cut - Dave sent me this blog; pretty amusing

DIE ANTWOORD - Afrikaans Anti-Heroes - awesome videos rave/rap semi-ironic trash - wow !

All the Strange Hours � Making and Thinking About Visual Art - I think his work is shite, but he writes some pretty interesting technical stuff about how artists look at color and the process of painting. I found him because I've been working on chroma and found these : flesh tones , chroma vs saturation , chroma

Munsell Color Science Lab FAQ is full of amusing stuff; it's not super technical, it's for lay people; interesting stuff on the eye and perception of color; this is also pretty amusing : Color Research Demonstration Images

02-05-10 - Car Dealers

Car dealers are such fucking cocks that they make me want to just walk away from the deal over and over. I swear they're like intentionally trying to make me not buy a car.

Some just intentionally lie to me to discourage me from being interested in cars. Like when I was looking at Caymans I was like "what about the 911?" and the guy was like "well that will run you $70k or more used, I think that's out of your range". WTF. I know it's more like $50k.

Some are just bizarrely unhelpful. Like when I drove the Mazdaspeed 3 and I was like "hmm I really like it but these seats are pretty awful, do you think I could swap them?" and the guy was just like "I don't think so, and it would be really expensive anyway". WTF.

Some are just super sleazy and weird. They love to ask you about what other cars you're considering and then tell you just absurd retarded negative things about them. Does that actually work on people? It just makes me want to get the fuck away from you.

In general they are just shockingly lazy and actually not pushy. I gave the Porsche dealer here a spec and told him if he could get a car like that in I'd probably pay their premium just to not have to deal with it. He can source cars from any other dealer around the country. Of course he makes no effort to actually do it, and I can see on the web searches that various matching cars exist.

Now the assholes I'm dealing with are just incredibly inflexible and dickish. They make me fight tooth and nail for every tiny concession. Jesus I'm trying my hardest to hand you a bunch of fucking money, do you want to sell or not? WTF. I'm so fucking tempted to walk away from the deal, but then I have to start all over again with someone else. URG.

Like for example, the inspector I sent found out the battery was dead, so I call and I'm like "yeah it looks okay but the battery is dead so you have to replace that" and they respond "I dunno if we can do any repairs at that price" ; WTF you're just being a dick for the sake of being a dick, we both know perfectly well that you will not sell me a car with a dead battery in it, and if I just complain a while you'll cave in, but you just want to add some extra unpleasantness like a shit cherry on this shit sundae.

I swear right now I'll never buy from a dealer again. If I want to buy used, I'll buy from an individual (and only the 1st owner). If I want to buy new I'll use one of those online car direct sites. It should be a nice experience buying an expensive car; it should be like flying first class. They should give you hot towels and champagne and call you "sir". Instead they stuff their shit-covered gangrenous cock in your face and yell "suck it" in your ear, and if you refuse then they don't let you buy a car.

2/04/2010

02-04-10 - Pit Bulls

A couple more recent pit bill attacks ( man killed on Jan 18 and girl mauled on Feb 2 ) should remind us all that it's fucking absurd that people have these dogs.

I'm not convinced that they should necessarily be illegal to own (or controlled, like owning a tiger is). But certainly the owners should be prosecuted for assault with a deadly weapon or negligent homicide or some shit like that.

The arguments that are made by the pro-pitbull lobby are so absurd. Like "they're really sweet dogs if you raise them right", "my dog is my companion how dare you outlaw it", blah blah - you know what? a Schnauzer is also a really sweet dog if you raise it right, and it could be your companion, and it won't bite off your baby's face. There's just absolutely no justification for why you would have that dog instead of a safer one.

This is also another one of those cases where I wish social scorn would step up. We shouldn't really need the government to intervene here. When someone says "hey wanna meet my pit bull?" you should respond with "keep that dangerous thing away from me, you deranged psychopath".

2/03/2010

02-03-10 - Car Pricing

I've noticed some interesting things looking at all these cars.

One funny thing is that *year* affects price a lot more than mileage. That is, a 2002 with 10,000 miles will generally be a lot cheaper than a 2004 with 30,000 miles. Now in some cases that makes sense because the older car has less features and is thus less desirable. But in many cases that makes no sense at all. The actual "age" of a car in terms of lifetime is its mileage. And car models tend to change discretely, not continuously from year to year. That is, there will be a major revision some year, and then it basically stays the same for many years. The reality of pricing is that people basically price by year and then just make a small adjustment for mileage; that's a big mistake which means you can exploit it by buying an early year car from a certain revision (though you usually don't want the first year of a revision).

For example with 911's there were major revisions in 1999, 2002, 2005, and 2009. Another thing I've noticed is that some year models are specifically known to be bad, but that is not particularly well reflected in the price, so you just have to avoid those years. So for example 2005 is known to be a lemon year. So you get maximum value by buying a low mileage 2002 or 2006. You get the worst value by buying a high mileage 2008 or 2004-05.

Another funny thing I've seen is if you search around a bit, you can often find the same car listed in multiple places with different prices. For example I saw a used car on a dealer's web site listed for $59k, then listed again on autotrader for $56k , then on ebay for $54k. You could probably get it for $52k if you fought hard; if you just saw the $59k on the dealer's site you might think you're getting a deal at $56k but in fact that's just their asking price on autotrader. In another case I saw a car on the Porsche.com dealer-swap network listed for $56k , on the dealer's own web site for $54k, and on autotrader for $52k.

Another useful thing I've found from searching is that you can use the google cache to see how long certain pages have been up, so you can tell how long a car has been advertised on the net. Obviously if it's been sitting for sale for many months you're gonna have a lot more opportunity to bargain.

I looked at one car at one of these luxury car consignment shops; the dealer guy there admitted to me that they don't really care that much about selling consigment cars, and he would try to get the owner to eat a low price to get it sold. Obviously the dealer is still trying to fuck you, but if he's also trying to fuck the owner then at least the fuckage is shared.

One cool thing about buying used cars in the modern age is they all have computer chips that record faults, so if you have them inspected by an independent mechanic who can pull the computer codes. The manufacturer uses these codes to see if you have done anything that would void warranty, so they're pretty good. It will tell you things like over-rev red line events, overheating events, low oil events, etc. Obviously the computer can be tampered with, but if it's been cleared you can see that too.

Something else I looked into a bit are the CPO (Certified Pre-owned) programs. They give you an extended warranty (usually 2-3 years) and perhaps more importantly they mean the car passes some checklist of basic problems, has decent tires & brake pads, etc. If you're buying a local car or a cheap car, this doesn't make any sense, but if you're buying a car that's far away, there's a pretty big bonus value to knowing it at least passes the CPO check before you fly out to see it yourself. Obviously extended warranties are always a bit of a ripoff, but say the CPO adds $3000 to the car price, the actual expected value of the warranty that you get back is maybe $2000, but the value of the inspection is $500 or so when you live far away, and the time savings is worth $500, so you wind out okay on the deal.

Oh, another thing that's useful to know is geographic variation is really severe. Long ago I wrote about the weird fact that used Subarus sell for so much money up here in WA that you can almost buy them in CA and drive them up here and make a career out of it. The same is roughly true of the 4WD Porsche 911 - it's around $10k cheaper in CA or AZ. Even aside from the 4WD / rain car variation, used cars are just way cheaper in the west/south, particularly in LA, Phoenix and Houston. They have way more selection down there, much bigger market, more competition.

2/01/2010

02-01-10 - Google Autocorrect

One of the fucking annoying things Google has silently added recently is auto-correcting searches. For example, try any of these searches :

prosac

forman

oper labs

and what you will get is searches that have results for :

prozac

foreman

opera labs

note that I'm not talking about the "did you mean" - I mean it silently decides to give you results for a different word. Awful.

Fortunately there is a trick to get around this. Just put your search word in quotes. Apparently you could also use a +, though that also makes the word mandatory.

See for example : here

It seems to me at least part of the problem is that the stemming is not aware of changing word meaning. eg. obviously if I search "widget" then results that contain "widgets" are probably good, but changing "prosac" to "prozac" is giving me completely unrelated results. You should only automatically broaden the search to word variants that are actually related.

old rants