Logan's World

McCain with 8

"It is McCain with an early lead of 8 electoral votes..." -- MSNBC.

Thanks guys, that's helpful. Way to call Kentucky.

KDE 4.1

I decided to finally give KDE 4.1 a try via the new Kubuntu 8.10. Unfortunately, it was a frustrating experience. In a matter of minutes I encountered numerous serious bugs, many of which were related to my multi-monitor setup. It appears that bug reports already exist for the issues I had (see below), so hopefully they'll be fixed soon.

My first issue: upon docking, I hit my Thinkpad's blue "ThinkVantage" button which I have set to run a really handy script to take care of everything I want done. It sets up the wired network, mounts my file server, and runs xrandr to get the multiple desktops set up. In KDE 4.1, kwin crashed. It does this reliably each and every time. I'm not sure how common this problem is, because it seems like one of those issues that if every multi-monitor user (via xrandr) encountered it, the developers would themselves be annoyed enough to fix it.

My workaround was pretty simple, and gave me some hope that I could work through the problems. I simply had my dock script kill plasma (the application which sets up the desktop taskbar, etc), enable the second monitor, and start plasma again. No more crashes; so far so good.

A few more minutes and the annoyances mounted. Plasma wouldn't draw the desktop background or have any "widgets" on my laptop monitor when in dual monitor mode. One annoying workaround here is to create a panel on the new monitor and restart KDE. Then on single-monitor mode, you have two panels.

This is amongst numerous crashes. Plasma would crash if I turned off the setting for view hidden files while searching for a background wallpaper. Kontact would crash fairly frequently, but I didn't spend much time seeing what was reproducible. One of my most-used features in 3.5 was in-line editing of the TODO list. This doesn't appear to be implemented in the KDE4 version of KDE-PIM.

I should mention the good: the desktop was beautiful, most of the applications I wanted have been ported (amarok will be soon). The direction of KDE looks very good; the default desktop setup is far superior in appearance and functionality to KDE 3.5. Desktop effects actually work, and aren't horribly slow. If I didn't value snappiness so much, I would actually use them.

In the end, though, the crashes and the broken multiple-desktop behavior was too much. Rather than spend more time trying to find workarounds, I looked some of the more annoying behavior to ensure that a bug report was already filed; then I installed KDE 3.5 and Kubuntu 8.04 again. KDE 4.1 has definite promise, but some things (like scrolling and text entry) were noticeably worse than 3.5. The frequent crashes of kontact, kwin, and plasma were really the deciding factor. I hate to say it (as a huge KDE fan), but 4.1 still feels like beta software, at least in my multi-monitor laptop config.

I'll give it another try in 6 months with Kubuntu 9.04 and KDE 4.2.

Plug-in hybrids and solar panels

I came across this article at AutoWeek today which presents the case against using solar panels on hybrid cars. What the article doesn't deal with, as at least one comment pointed out, is that few people realistically expect the car to power itself in real time with solar energy. The real question is, would a reasonable solar panel roof produce enough electricity to charge a plug-in hybrid's batteries while parked?

This back-of-the envelope calculation requires a few pieces of data. First, how much potential solar energy is there for most people? Fortunately, cars can be readily parked in direct sunlight if desired, and people tend to park their cars at work in parking lots. This obviously doesn't work for people who park in garages, but you could certainly install a modest solar array on top of your garage (while adding one to your car would almost certainly be more expensive). Still, in the spirit of the AutoWeek article let's stick with the solar panel on car roof idea.

Lots of places provide rather dramatically varying estimates of potential solar energy (I emphasize potential because photovoltaic cells are roughly 10-15% efficient). Of course, worldwide or national averages are not terribly useful as the solar energy varies tremendously by region. To give some idea, check out this map. The second link assumes a southern tilt to the panel, which would have to be accommodated in a car with either a tilting panel on the roof or a built-in tilt and parking the car in the right orientation (more info here). Keep in mind that the measure in those is kilowatt hours per square meter per day. The definition of "day" of course, also varies by season and latitude.

To make life a little easier, assume that the owner of a plug-in hybrid with a solar roof chooses to keep the car outside during days with good weather. The estimates on the map appear to range from about 4kWh to 6.5 kWh per day per square meter. Since photovoltaic cells are roughly 10-15% efficient (depending on the technology used, which also affect costs substantially), this gives a range of 400 Wh - 975 Wh per square meter per day. Cars aren't very big, and many people wouldn't want photo cells plastered over the trunk or hood, so I'll consider 1-2 square meters of photo cells on a car. That makes the range 400 Wh/day on the low end to 1950 Wh/day on high end. A dramatic range, but considering the vastly different environments that people even within the U.S. have, it seems appropriate. A possibly important assumption I'll make is that the charging efficiency of plug-in is 82%, which is a number that comes from charging with a 120V (standard U.S. outlet) source. That provides the practical energy range of 328 Wh - 1600 Wh stored in the batteries (assuming the batteries aren't already at full charge, of course).

The other bit of data is how much energy the Prius (or similar electrically-capable) car needs. I've found a number of not terribly reliable sources claim about 260 Wh/mile for a Prius and this source indicates that hybrids in electric mode take between 180-460 Wh/mile, which is in line with the Prius numbers I'm seeing. Put that together with the solar energy stored and you get 0.7 - 8.9 miles depending on all of the factors mentioned above.

The question posed here is not "is putting a solar panel on your car roof going to save you money?" but rather "is putting a solar panel on your car roof going to have anything like a reasonable effect on your gas mileage/power bill?". At the low end, assuming a relatively energy-guzzling hybrid, low photo cell efficiency, lack of sunlight and a 1 square meter panel on your car, you get 0.7 miles. That's not going to put a dent in even a modest commute (if it did, you'd be better off walking in all likelihood). At the more optimistic end, you get 8.9 miles. Doesn't sound bad in ideal conditions, depending of course on how much such a system costs. If you end up using all of the solar energy in a given week, you'll get 62 miles, or roughly 1-1.5 gallons of gas per week (depending on how fuel efficient your plug-in hybrid is as using gas). Over a year, even at $5/gallon gas, it is $390 at the most optimistic. Again, I can't answer the question about whether the solar panel on the roof will save money, because any such system now would involve custom modifications which would almost certainly not be the cost-saving choice. Whether Toyota or other companies can pull off panels which are cheap enough is another matter.

It doesn't take too much searching to figure out that somebody's already tried this with a 2001 model Prius. Their numbers seemed optimistic to me - taking only summer months in Ontario as their benchmark. Still, it's a real-world exercise, so how did it stack up to the calculations above? Their theoretical claim was 8 km (about 5 miles) assuming 6 kWh of energy in Ontario sun, 2 square meters, and 10% efficiency. They assume 150Wh/km, which translates to 241 Wh/mile. They also assume no loss from charging. Their practical results were in terms that are hard to compare - a 10% fuel efficiency improvement, but with what sort of usage? If you only commute 2.5 miles each way every day, you could conceivably use no gas at all.

So the bottom line is, if you live in an area in the right conditions, drive a small enough hybrid (Prius or smaller), and commute a relatively short distance, you could see some significant gas savings with 2 square meters of panel on your car. Is it likely to be worthwhile for many people? Not unless the efficiency of the panels goes way up and/or the cost becomes quite affordable. If either happens, installing panels as optional equipment on hybrids could become common, but I don't see it happening on the manufacturer level at this point. Besides, to my knowledge commercial hybrids aren't even plug in yet.

A more practical solution is to have solar panels on the roof of your house/garage. These are less custom and becoming more widely adopted, so installation and per square meter costs should continue to improve. A house is in a fixed location, so the solar panels can be installed at an ideal angle to get the most sunlight. Finally, a house is almost certainly able to make use of the excess energy through appliances like a refrigerator, lights, etc. If not, in most places you can sell excess energy to the power company back on the grid, lowering your energy bill.

TaxCut 2007 for Mac

Having used TaxCut for the last few years, I've been pretty happy with it overall. This year, however, I ran into a really annoying bug with the Mac version. I bought the Federal + State version retail and tried to install my state, but TaxCut wanted to charge me $29.95. It seems that the Mac version is blissfully unaware that I paid for the combined version, even though the box, CD, and CD cover are all clearly marked as including the state (not to mention the higher retail price).

Apparently I'm not the only one, as a few quick Google searches turned up people with the same problem. Rather than waiting on hold to try to explain this over the phone, I did the TaxCut online chat. The first guy I got seemed to have trouble understanding my problem, but after the chat disconnected (it was an auto-refreshing web page that started giving me a 404 not found error), I connected to a different guy. It took maybe 10 minutes of waiting for a representative but only about 2 minutes for him to issue me a download for my state.

Nonetheless, I'm not impressed by TaxCut's QA this year. One would think they'd have patched this issue, but the very first thing I always do upon installing TaxCut is to update the software. My other thought was that maybe it was getting confused by TaxCut 2006's "Library" (settings) files, but moving those out of the way didn't change anything. As far as I know, there is no workaround short of contacting tech support. Hopefully they'll patch things soon, because I imagine the do-it-yourself tax preparers really get into gear in March.

Krystaline icon set for KDE

My favorite icon set for KDE is Krystaline. When I went to download it for my desktop, the link was broken. I happened to have a copy left on my laptop. Fortunately, it's licensed under the LGPL, so I can mirror it here.

Food miles and true environmentalism

An op-ed in today's NYTimes details the economic arguments surrounding "food miles", or using the number of miles from where the food was produced to where it is consumed as an indicator of how friendly the food is regarding global warming. The thought process goes that transportation of food involves significant carbon-dioxide producing shipment methods. Thus, the thought process goes, to reduce carbon dioxide emissions, one should eat from sources closer to home.

As with so many things in environmentalism, that is simply the first layer. The op-ed correctly points out that carbon dioxide emissions from transportation form simply one part of the overall production process. Buying local might mean buying with inefficient production methods - more watering, greenhouses, fertilizer, etc, if the food you're buying isn't well adapted to the current climate. The article uses the example of New Zealand lamb versus UK lamb, and that even transporting lamb across the world by boat it is still more environmentally friendly to buy the NZ lamb.

So the global warming-concerned buyer is left with a problem: this "food miles" number may not actually mean more carbon-friendly. What's a solution? Some economists have long argued that the efficient way to reduce carbon emissions is by taxation. Carbon emissions are an externality, creating a market failure in which buyers of goods which involve a great deal of carbon emissions are not paying the social cost. A carbon tax will increase the price of relatively polluting products compared to those which are environmentally friendly. The price of a good - say, NZ lamb versus UK lamb, would in part reflect the environmental benefits of the NZ lamb. This tax would induce companies to find ways to reduce carbon emissions through better technology to compete in their markets. The customer can go back to evaluating products on an even basis: by their prices.

Whether to hire a real estate agent

The New York Times has an interesting article about for-sale-by-owner (FSBO) versus real estate agent home sales. The classic trade-off is as follows: a real estate agent may help sell a home faster, for a higher price, and with less work than selling yourself. The downside, of course, is the percentage that real estate agents take off the sale price (6 percent is common). Data for Madison, WI properties were analyzed by two economists at Northwestern and an economist at UW: Igal Hendel, Francois Ortalo-Magne, and Aviv Nevo. The punchline is that for Madison properties, FSBO did about as well as properties sold by real estate agents. Factor in the agent fee, and it makes sense to list yourself.

Of course, it's useful to keep in mind the incentives of real estate agents. The book Freakonomics by Steven Levitt and Stephen Dubner highlights this in their second chapter. Realtors definitely have an incentive to sell quickly, to move their efforts on to the next house. Unfortunately, diminishing returns may hit realtors' incentives hard. Getting an extra $20,000 on a $200,000 home may take more time than selling the house quickly and devoting that time to selling more homes.

One thing about housing prices is that it's not trivial to compare one house sale to another. So many unique characteristics make up the sale price of a home that it's hard to control for them. If you're not careful, you won't sufficiently control for selection bias; for example, suppose more expensive houses tend to be sold by owner. One hypothesis might be that these will tend not to be first time home sellers, and thus more comfortable handling a sale without a realtor. If you don't control very carefully for house quality, it'll seem that FSBO houses sell for more than realtor houses. Of course, with the statistics that the realtor associations cite in the article, the reverse might also be the case.

The paper is available in working paper form, interestingly enough, from the NYTimes itself. Strangely, the NYTimes leaves out the third co-author in the article: Francois Ortalo-Magne. I suspect it was in an attempt to frame it around the other two economists' argument, but it seems poor to leave out a third co-author.