Sunday, March 27, 2011

Ubuntu has not solved the agency problem in community software development for the Linux desktop

Four and a half years ago, I wrote:

In an open source project with a dictatorial or committee-led governance structure, somebody would long ago have cracked some heads and gotten this feature implemented. In a commercial software project, open source or non-, some engineer would be assigned ownership of this feature; and goddammit, if that feature didn't get implemented and maintained, that engineer would be fired and the feature would be assigned to someone else. But KDE's headless. It's less like a mammal with a central nervous system than an enormous amoeba whose various pseudopodia ooze tropically in the direction of "developer itches" and "coolest implementation hacks" (hence the recent proliferation of "hugely ambitious infrastructure refactoring" subprojects like Plasma or Solid) rather than unsexy, annoying-to-implement features that merely provide value to end users.

I genuinely thought Ubuntu had a fighting chance of resolving this agency problem. Surely with a dictator taking responsibility for the entire desktop stack, there would be progress. When there's a bad corner of usability for a common user task, somebody will crack some heads and get it fixed.

Today, I tried to share a file over my local network between my Ubuntu desktop and my Ubuntu laptop. And I ran into this defect which has been open for a year. In the year 2011, the easiest way to copy a file from one Linux computer on your local network to another Linux computer on your local network is still either (a) copy it onto a USB stick or (b) upload it to the Internet (e.g. by attaching it to a draft Gmail message).

Meanwhile, Ubuntu's burning huge numbers of developer and UI designer cycles on stuff like Unity ("We're too impatient to fix the rough edges in our existing desktop which has a decade of developer investment behind it; therefore we will design a brand new desktop, because there definitely won't be any rough edges in that.").

It turns out that, in fact, Ubuntu has not solved the agency problem in Linux desktop software development. Sigh.

Thursday, March 24, 2011

Federal budgets and applied public choice theory

Atrios writes something so true you should affix it in your memory for the next two decades:

There's literally nothing that this Congress today can do to reduce the deficit 20 years from now. What you can do is sign into law legislation which reduces granny's pension 20 years from now. And, yes, given the way our system works it wouldn't necessarily be easy to reverse that decision 20 years from now depending on the politics and who is in power. But what will still be easy to do 20 years from now is cutting taxes on rich people and writing giant checks to defense contractors. Those things are always easy to do when Congress and their donors are mostly rich people. Reduce the deficit by cutting granny's pension, increase it again by cutting taxes on rich people. Rinse repeat.

. . .

. . . just to remind us of history I'm sure we all remember. That Democratic Socialist Bill Clinton got rid of the deficit. Alan Greenspan, who spent years fretting about the deficit, suddenly decided the great danger we faced was not having a deficit. And Bush tax cuts, and too and such.

Once this really sinks in, you realize that the only measure which leads to long-term balanced budgets is reform which changes the configuration of political power so that budget deficits no longer benefit the powerful. Going by today's projections, the main contributors to long-term budget deficits are rising medical costs, the Bush tax cuts, and war (and preparation for war). Therefore the principal budget-balancing methods that might work in the long term are:

  1. health care reform that reduces the power of people and industries that make medical care expensive.
  2. campaign finance reform and progressive taxation, which reduces the power of rich people to demand tax cuts (campaign finance reform directly, and progressive taxation indirectly by simply making rich people less rich).
  3. reducing funding for the military and defense contractors, which reduces the political power of the military-industrial complex by reducing the number of people dependent upon it.

You may think some of these changes would be bad. That's OK; it just means these reforms conflict with your political values, and I'm not even trying to argue you out of your political values. Just recognize, then, that budget deficits will forever be the price of maintaining policies consistent with your political values. Relax; it's not so bad; maybe it's even a price worth paying; but no amount of cleverness on your part will allow you to wriggle out of paying it.

Your faction may exercise heroic effort, tenacity, and ingenuity to bring the budget into balance. And if you succeed, the current political economy of the United States simply means that the politically powerful health care sector, the politically powerful overclass of wealthy people, or the politically powerful military-industrial complex will find some way to squander your effort and throw the government back into deficit, with the balance of money going into their pockets. This is not a piece of political polemic; it is simply an observation that I offer about the world, with recent history as my evidence.

And as for the widespread conceit of upper-middle-class liberals that there's some purely technocratic way to fix long-term deficit problems by twiddling around with the retirement age and the like — that's just an exercise in hopeless naivete.

Sunday, March 13, 2011

The moral case for sending money to Japan (rather than somewhere else)

Japan is one of the richest countries in the world, both in absolute and per capita terms. The earthquake and tsunami have inflicted terrible damage on the country, leading most likely to thousands of deaths and many millions of dollars of property damage. However, as a nation, Japan has ample financial resources to recover from the disaster. Japan was a healthy and prosperous society one week ago, and a year from now the overwhelming majority of Japan's people will still be alive and healthy, and they will still be relatively prosperous in global terms.

By contrast, there are still hundreds of millions of other people around the globe living in conditions of persistent poverty and immiseration. Just to take one random example, one year after Haiti's 2010 earthquake, Haiti is still a complete basket case.

So, the Japanese earthquake has no doubt pricked your conscience. You have been reminded that there are people in faraway places who direly need assistance. Your moral intuitions are worthy, but if they lead you to donate money to Japanese relief, then you are probably doing something non-optimal from the point of view of improving human welfare. Donate, instead, to an international relief organization that consistently directs its efforts to the most needy worldwide: Oxfam International, Unicef, etc. It is even possible that these organizations will spend some of their resources to help Japan now; but they're in a much better position to analyze the situation and direct the appropriate quantity of resources in that direction than you are.

Accordingly, I just donated to Oxfam, and I urge you to do the same.

On the other hand, of course there are some forms of aid specific to the immediate aftermath of natural disasters for which money is not a substitute. Obama has directed the U.S. Navy to station aircraft carriers in Japan to help airlift relief supplies and such. Google has launched the People Finder for Japan. Etc. These organizations are uniquely situated to help in ways that no amount of money can purchase on the open market. And if you know of some similarly specific aid effort for which equivalents cannot be obtained via market mechanisms, then you should support that effort however you can.

A final caveat is that resources within Japan are unequally distributed. There are usually some very poor and miserable people even within rich societies. If you know of some specific subgroup within Japan which is unlikely to receive assistance due to the structure of Japanese society, then again go ahead and donate to help them as well.

But to a first approximation, the logical response to natural disasters in wealthy countries is to donate money to aid organizations generally, not to donate to aid for those countries. And yes, when The Big One hits the Bay Area (where I currently live), I'll say the same thing.


UPDATE 2011-03-16: Via MR, charity ratings organization GiveWell agrees that donations should not be sent to funds earmarked for Japanese disaster relief. See also F. Salmon, and F. Salmon again.

Tuesday, February 15, 2011

A simple handwave that makes The Matrix tolerable for the scientifically literate

Suppose that controlled fusion power requires real-time control computations which can be much more efficiently implemented in neurological hardware than in silicon.

(That Second Law of Thermodynamics thing was bothering the hell out of you, wasn't it?)

Note that neurological hardware really is exceptionally power-efficient for certain classes of computations. Common estimates of the human brain's power consumption are 20-25 watts; this is roughly the wattage of a Mobile Intel Core i5 processor, which (as far as we know) appears to be a much less powerful computer for many purposes. By contrast Watson runs on 90 IBM Power750 servers, filling ten racks, whose power draw is something like 80 kilowatts. In other words, Watson consumed about four thousand times more power than either of its meat-based competitors.

Here in reality, I think it's unlikely that any fixed class of computations can be efficiently implemented in neurons but not in silicon — see Carver Mead and his academic descendants' work on analog silicon circuits. But positing that such computations may exist seems within the realm of acceptable science-fictional handwaving.


UPDATE: Yes this is close to the standard handwave that the humans are being kept as computing devices, not power sources. But I think you need to draw the connection explicitly to power generation; otherwise there's just too much narrative in the film and animated shorts that makes no sense.

UPDATE': Never mind, I just remembered Morpheus's exact wording from the first film's voice over and I don't think it's salvageable. Oh well.

Sunday, February 06, 2011

Why iPads' role in education should be limited

This morning I came across this story about a Georgia state senator proposing replacing textbooks with iPads. This strikes me as a bad idea for several reasons.

First, the iOs developer agreement means that iOs will forever be a platform that teaches children to consume rather than to create, at least with respect to my discipline (computer programming).

If you think this doesn't matter because "kids don't program", I would like to disabuse you of that notion, hard. The iPad could have been a great device for running Squeak, which includes environments like etoys and scratch. These systems teach children to be producers rather than merely consumers of computing, which I think is no small matter given the importance of computers in society today. But, as a programming language interpreter and compiler, Squeak is prohibited by the iOs developer agreement

Many adults grew up as passive consumers of technology — for example, as television watchers or video game players, rather than people who shoot videos or program games. This blinds them to the possibility that things could be profoundly different. Ubiquitous digital video cameras and video sharing sites like YouTube have already transformed the relationship between people (especially young people) and video. Video is no longer solely the thing you sit on the couch and watch. It's something everyone can and does create. Despite the proliferation of astonishing banality on YouTube, I claim that this is a positive development. Making and editing a video of your cat is still more rewarding, and exercises more cognitive skills, than passively watching almost anything on television.* Programming could be the same way. No, the average person will never build complex software from the ground up. But the average person can, with appropriate support, learn to write small programs, and to modify big programs around the edges, and (crucially) to enjoy doing so. The iOs developer agreement simply shuts down this avenue of creativity, and that's the wrong thing to be doing to children.

Well, OK, so maybe you don't care about ideological issues like "freedom" and "creativity" (although I claim that if you do not, then you have no business working in education). Let's get down to practicalities.

The general attitude of Apple towards iOs device "owners" has been that the end-user does not truly own the device — Steve is just letting you hold it for a while. Note Apple's hostility to jailbroken devices and even its use of nonstandard screws for its cases. What does this mean in the education world? A school district that buys a thousand iPads will be completely at Apple's mercy w.r.t. software upgrades, hardware repairs, and general system maintenance.

The article notes that Georgia is "currently spending about $40 million a year on books [that] last about seven years". Apple is a consumer company and builds its products to last about 2 years. Ask people with vintage 2007 iPhones how well their devices are coping with iOs 4. Furthermore this 2-year product cycle is not an accident; it's built into the foundation of Apple's current business model, which is to ride the leading edge of technology so that they can always sell a device that's shinier than their competitors' devices (and even their own devices from a couple of years ago). Have the school districts really worked out the implications of this? I doubt it.

Furthermore, the article notes: "Textbook publishers are eyeing the potential for moving their content to the digital world, enabling them to update material rapidly and include interactivity." The unspoken subtext is that school districts will move from purchasing textbooks to purchasing subscriptions to textbook content. Ask university libraries with electronic journal subscriptions how well that's working out for them. And the iPad textbook age will, in the long run, be worse than that. Electronic journals at least distribute their articles as DRM-free PDFs: if you find a paper you need, you can save it to your local hard drive and have it forever. Does anyone think that grade school textbook publishers are going to release their iPad content without DRM? I laugh mordantly in your general direction. I pity the school district that has a budget crisis and cannot afford a textbook publisher's access fees for the year. I can only hope that most districts keep the old dead-tree textbooks in the basement for emergencies.

In summary, iPads in education will

  • Teach children to consume programs rather than to create them.
  • Lock schools into a closed computing ecosystem that they do not control.
  • Lock schools into a continual treadmill of costly hardware upgrades.
  • Lock schools into electronic subscriptions to DRM-encumbered textbook content.

All that said, dead tree textbooks are expensive and heavy and troublesome to update, and there's tremendous potential for electronic educational materials to improve the situation. And the iPad, being an excellently made device with an active developer community, is a fruitful context in which to experiment with new educational software. But until/unless some of the above features change, it's extremely premature for an entire state to make the iPad the center of its educational curriculum.


*Of course, there are a few great shows on television and people still watch those. Homemade video does not replace that and never will. But great shows constitute a tiny fraction of televisual content.