Archives for: March 2005


prepare for a life of suffering my little anime dork

posted by ben on 05.03.31 at 22:53, null, null, Leave a comment Permalink
Are those tentacles?
No, it's a battery pack coming up around her back.
Does she have charge punch?
Like in Spirited Away when she...
(pause to sigh appreciatively)
There's not much shading on her legs.
There's shading, it's not detailed.
What's the difference?
When I shade her shoes that's detail.
I want to go see Sin City tommorow.
Why didn't you go today?
It didn't feel right.
After eight. Eight is the Battlestar Galactica finale.
I missed one.
I did too.
I didn't know it was the doctor.
The doctor?
He recommended Ishmael.
Why don't you ever read anything I recommend?
He said Snowcrash was the best book ever.
And it was.
Why don't you read anything I recommend?

The dangerous world of pioneering...

posted by anwar on 05.03.31 at 22:37, Engineering, null, Leave a comment Permalink

Looks like the end of days for transmeta. They had a nifty idea with the realtime dynamic binary translation, but in the end - it didn't save them enough transistors to make a difference.

It will be sad to see them go-- they did the industry a great favor by showing consumers (and chip architects) that power mattered. Its too bad they didn't get more $monetary$ benefit from their efforts, before the mighty manufacturing machine at Intel turned a couple knobs and dusted off an old design and built it in the new process (Ah! The miracles of voltage scaling).

But then again, I'd rather be the lead dog any day of the week, he's the only one with the great view.

posted by collin on 05.03.31 at 21:51, nonsense, null, 2 comments Permalink

It's like reading a bottle of Dr. Bronner's soap, if Dr. Bronner was a "mathematician" that didn't get into Caltech. "Obscurantism?" That's almost as good as "symbological" or "technicated."

There is proof that 3 dimensional
math is erroneous, and that linear
Time is actually of a Cubic nature.
Ignoring Cubium indicts you evil.

Scholars ravage Earth's resources.
Scholars plunder America's Nature.
Nevada's 77,000 ton nuclear waste
burial should be made a monument
to Scholasticism's evil singularity -
contradicted by Cubic Created life,
relative to the Cubium dipole atom.

Time Cube is "T.O.E.", theory of everything.
Time, Life and Truth = a Cubic Principle,
a natural creation of ineffable opposites.
Caltech professors practice obscurantism
and can not ever allow Time Cube Debate -
for it will indict the word bastards as evil.
They can't argue it, their power is to ignore -
and that's why students must demand debate.
Word is most efficient form of enslavement.
You have been educated singularity stupid.
You have been educated in singularity evil.
Self singularity is cursed form of humanity.
Do you care to know that Cubic Creation
debunks the Word God and Word World of
the educated stupid human Word Animal?

comment by ben on 05.03.31 at 23:04
"You are a personified pyramid corner."
comment by collin on 05.04.02 at 11:23
Crap, symbological is an actual word...

Symbological \Sym`bo*log"i*cal\, a.
Pertaining to a symbology; versed in, or characterized by,symbology.
[1913 Webster]

algebraic properties of blogs

posted by ben on 05.03.31 at 20:36, null, null, Leave a comment Permalink

Where p is a post and c a comment. Which is correct? Are they both? Are they incompatible?


posted by collin on 05.03.31 at 16:54, null, null, Leave a comment Permalink

I was just listening to NPR and there was something about the latest election in Zimbabwe. One thing that was mentioned that I didn't know was that it is illegal to sell cornmeal, cornmeal being the main staple. Only the government is allowed to sell/distribute cornmeal and routinely arrests people selling cornmeal at roadside markets to make some cash, and sets up road blocks to prevent... What? Trafficing in fucking cornmeal? As one might imagine this doesn't work very well. Which has lead to at least one slogan for the opposition party, "Vote with your stomach."

We have issues.

posted by ben on 05.03.31 at 15:57, null, null, 9 comments Permalink

I think this also points out that some amount of education is always going to be required, although it may become small enough that people absorb it more or less naturally in the course of growing up.


In Birds Without Wings, de Bernieres talks about how difficult it was to use a car in 1920 or so. Now even my little cousins can drive cars... terrifying but true. 80 years ago you needed a chauffeur/mechanic, untold numbers of replacement parts, spare gas, and a knowledge of how to use all those things. Now cars are automatic with power steering. Before only the extremely wealthy and knowledgeable could have a car. Now virtually everyone has one.

I say this is because the technology has become intuitive. The exposed UI is simple, easy to use and functional. The hideous underlayer is rarely exposed. This is what people want. This is what I want.

comment by collin on 05.03.31 at 16:48
Posting an image of a 1930 Jaguar fits well with your point. But unless I'm mistaken, by 1920 people could have purchased a Model T which is a very different car. I would venture a guess that the very wealthy had chauffeurs and mechanics not because of the inherent difficulties in the use and maintainence of a car, but instead that having large amounts of money allowed them to pay others to perfom these tasks (ignoring social expectations), which is the point you were making previously. Regardless, good UI is a good thing, unless other things are more important. Correct me if I'm wrong, but you don't drive a 2000 Honda Civic. You drive a 1975 911 Targa. This isn't because the technology is worse now than it was 25+ years ago. Driving an old Porsche IS more difficult than a new Honda, driving the former requires more skill because it's rear-wheel drive, is a stick, and has no power steering or ABS. I can't say why you drive what you do, though I have an idea: you are willing to sacrifice ease of use for quality of use. I know that's why I have/had a Porsche. And I know that I drive old trucks because sometimes when you sacrifice ease of use for simple engineering fewer things break and when they do break they are easier to fix.
comment by marco on 05.03.31 at 16:57
And on the subject of being easier or harder to fix, cars today are going in quite a different direction. Gone are the days when you could fix up your car if you wanted to, and nearly gone are the days when you could take your car to the corner mechanic to get it fixed. Now cars have opaque computers with hidden technical codes controlling everything, and only the manufacturer knows those codes, so if your car breaks down you have no choice but to take it to the dealer, who can then charge you whatever they want to fix it.

That's generally the same direction Microsoft (and Sun, and IBM) wants to take your computer in too: you pay MS lots of money, they provide you with a computer that does what you want, more or less. If it breaks or you can't get it to do what you want, you pay them more money to get it to work again, without ever having (or being able) to know how it really works. That's fine if you have lots of money and don't want to know how it works. That sucks if you want to know how it works and/or don't want to give lots of money to MS and company.
comment by devin on 05.03.31 at 17:18
I say this is because the technology (cars) has become intuitive. The exposed UI is simple, easy to use and functional. The hideous underlayer is rarely exposed. This is what people want. This is what I want.

Actually, this is not true. It is a very good example of bad UI that seems intuitive because it is familiar. But, in fact, the hideous underlayer is consistently exposed, much to the interface's detriment. Consider the following issues:

  1. Shifter — The gears in manual cars are still laid out for the benefit of the transmission, not for the benefit of the user. If we must have manual transmissions (we don't really need them in most cases, after all), they should be ordered according from low to high.
  2. Pedals — Similarly, it is really silly to use the same foot for both acceleration and braking. This is awful design — imagine how much quicker we could react if we used separate feet, or even better, our much more responsive hands? This kills people, but for historical reasons we accept it. In manual cars, using a pedal for the clutch doesn't make sense — with modern hydraulics we shouldn't need the strength of our feet to push the clutch in.
  3. Steering wheel — Again, for historical reasons, the steering wheel turns the wheels, not the car. If a UI designer had designed the car, turning the wheel would turn the car — when you stopped turning the wheel, the car would stop turning — or better yet, we'd use a joystick — we only have such monstrous steering wheels because in the old days before power steering you needed the torque. In any case, the mapping from a steering wheel to a turning car is all wrong. This is most apparent when you are backing up — you have to turn the wheel the opposite of the way that you would usually would it in order to turn in the same direction. I certainly back up the wrong way for a bit when I'm tired — I've seen other people do this too.
  4. Turn Signals — On most cars you push something down to turn on your left signal and push something up to turn on your right signal. Again, the violates the UI precept that the mappings should correspond to the user's understanding of them.

And we've all had issues when trying to get an unfamiliar car to work. This is because 1) there aren't strong interface standards and 2) the current UIs don't offer good affordances. We have to push, pull, and prod every dangling bit just to figure out how to spray the windshield cleaner. We have to examine the top of the shifter, we have to test out how finicky the clutch is.

In all of these cases, we have forced the user to conform to the system image (i.e. the car's mechanics) rather than create clean, well-mapped interface with good affordances that allow faster reaction times. It is wrong to confuse familiar with intuitive. This is the first thing they taught us in HCI.

comment by ben on 05.03.31 at 18:13


  1. Shifter — Cars are automatic now. The exceptions are sports cars and some hybrids. I assume the hybrid thing is some electric/gas issue and will go away. Race cars put the gears in a line, but I doubt consumer cars will ever see this because consumer cars are automatic.
  2. Pedals — I've read about combined gas/brake pedals from decades ago. I think the problem is that the pedals are tightly modulated resulting in accidental slamming on of the brakes. I would also imagine you might wear out the pads more quickly while attempting to coast. The picture is from a contemporary version of this.
    I know you don't want to use a gas and brake pedal with each foot. I tried this while learning to drive and the result was simultaneous gas and brake. The one foot thing prevents this.
  3. Steering wheel — The way the steering wheel works makes sense to me. Car video games are very hard for me. Press the button, turn a little, then it stops... then my little cousin slams my GT2 into a barrier and mocks me.
  4. Turn Signals — Do you really want the turn signal sticking out the top of the steering column? I like it on the steering column and on top is the only way I can think of to get left/right movement.
comment by anwar on 05.03.31 at 23:03
Devin and Ben are both right here. Cars, computers, and most other *young* tools have (sometimes monstrous) UI deficiencies. This is not entirely bad, as it leaves some interesting design work for us! (what would engineers do in a world with 100% complete tools?)

So far I've spoken about Cars and computers as tools (you see - I drive an (unmodified) Honda Civic aka a tool which takes me from "here" to "there") and at some point, the tool is "good enough" -- for a certain cost (e.g. design effort, money, time) it performs the task better than the alternatives (including not using the tool).

As to some insight from the *other* perspective on cars...

I remember Ben telling me about his Porsche, "Its like an extension of myself" in that he could make the car carry out his will. But his car is also a tool, differing from mine in that it not only gets him from here to there, it seems to be a tool in his quest for self-actualization.

And that is a great reason to try and improve this particular "good enough" device.

comment by anwar on 05.03.31 at 23:28

I have heard about the attempt to keep diagnostic codes secret by some automakers, but I think that there is an important distinction to make here. Decreased "fixability" due to having completely opaque systems is *different* from the decreased fixability due to increased integration.

We should do all we can to ensure that we buy "open" designs (or at least make sure that reverse-engineering is always legal). But integration is not always a bad thing.

One trend I do see with cars is that (at least in developed countries) parts are cheaper than a mechanic's time. This is why we just install a new muffler instead of just brazing a small piece of metal over a hole in the muffler (which they do where labor is cheap). As well, increased levels of integration have led to increased performance, reliability, safety, and lowered cost**.

**So this lowered cost is an interesting issue, because whether or not this is a GoodThing, depends on whose costs are being reduced. If every function of your car can be integrated onto one widget, and this leads to a $5000 repair bill when the widget dies, it is probably not so good for you the consumer. Then again, car manufacturers have a vested interest not to do this, because for non-emotional car purchases, this increased TCO would be a serious detriment. (lets assume that we have rational people purchasing these cars, not "consumers" for a moment)

comment by devin on 05.04.01 at 01:19
Your points illustrate exactly why the UIs for cars are so bad -- the exposed interface is determined solely (or mostly) by the physical exigencies of the car. That's like saying apache can't have a good UI because it is a daemon.

Saying, for example, that we can't use different feet for the gas and brake pedal because we'll use them both at once, or wear on the brake pads, is bogus. We could easily design a system where this didn't matter; we could make it so that when the brake is pressed, the gas stops. Or we could control the gas and the brake with one hand -- forward to accelerate, backward to to brake. The point is, it's usually better to spend the time to give it a good UI and engineer around the problems this creates, than to give it a crappy UI because a good one doesn't match one-to-one match with the internals.

Similarly, the reason that turn signal's left and right are up and down is precisely because they are on the steering column. Why in god's name do we have a steering column? In modern cars they aren't connected directly the rest of the car anyway (kills people on front impact). We should be trying to figure out a configuration where left and right map to left and right rather than coming up with excuses as to why we can't.

These are both examples of where the internals of the machine have dictated the external interface. This is not how good UI is created. This is 2005. It's time to stop making excuses. We should design for the user, not the machine.

You write: "The way the steering wheel works makes sense to me. Car video games are very hard for me. Press the button, turn a little, then it stops... then my little cousin slams my GT2 into a barrier and mocks me."

It's good that it makes sense to you. This is bare minimum requirement for a UI -- that the user can come up with a model that allows him to predict what will happen if he does something. If not not, he'd have to do everything by trial and error.

But again, you're confusing good UI with familiar UI. This is like asserting QWERTY is better than Dvorak because you're faster at it. This kind of sentiment is, as you say, why open source UI's suck so much.
comment by ben on 05.04.01 at 12:40

I just realized... I think tiptronic is basically what you want from shifting. It's automatic, but you can also shift in a line.

The problem with early power steering was that there was no feedback. That well dressed woman in the photo might well be destroying a small European car as she backs up, but her Cadillac has such buttery smooth steering she can't feel a thing. That's precisely what I'd worry about from a joystick... (and playing games with a feedback joystick is hard... then it's my mech that explodes, not my car).

If a steering wheel worked the way you suggest, I'd have to turn it and turn it, round and round to make a gradual turn on a highway. I like turning it a little bit.

I don't think it makes sense to tie the brake and gas. They do two very different things. The gas provides actual acceleration. The brake only results in accelaration. Locking the brakes does not cause anything to spin slower and slower, it merely stops them. Even with ABS, right? That might make sense...

I also think there's a difference between software UI and engineering. Cars actually have to relate to the real world and they explode when they don't. With computers the connection is more tenuous. We are free to do much stranger things with UI in computers. The limits are what we want to do, whereas with a car the limits are more speed limits and the purpose of going somewhere and so forth. Software has freedom mechanical engineering doesn't.

Check it yo.

comment by ben on 05.04.01 at 12:55

I don't really want a UI-theory based car interface or book interface. God help us if these mythic 15 UI-theory people get a hand on our art.

If the theory is sound, then a UI-theory for software might be ok. However, the theory people have strange ideas no matter what field they practice theory in, they need to in order to justify their continued existence.

Engineers with common sense and a grasp of what people want do better than any theory consortium. This is a reason engineers should not merely be trained in whatever dark art they might practice, but in the world around them as well.

A question of etiquette...

posted by collin on 05.03.31 at 15:45, null, null, Leave a comment Permalink

Suppose you live in a apartment building. You walk down stairs, and in the foyer of your building you see two people; a homeless mute Vietnamese man with five teeth wearing a studded leather jacket and an albino Chicano gangbanger type. They are smoking a blunt and offer you a drag. Do you partake?

Old fashioned morphine...

posted by collin on 05.03.31 at 15:39, music, null, Leave a comment Permalink

"Old Fashioned Morphine"
by Jolie Holland. It's a riff on the old classic "Old time Religion," think somewhere between Billie Holiday and east Texas Blues.

Gimme that old fashioned morphine.
Gimme that old fashioned morphine.
Gimme that old fashioned morphine.
It's good enough for me.

It was good enough for Billy Boroughs.
It was good enough for Billy Boroughs.
It was good enough for Billy Boroughs.
It's good enough for me.

Sister don't get worried.
Sister don't get worried.
Sister don't get worried.
Casue the world is almost done.

So that's where it came from

posted by marco on 05.03.31 at 13:18, null, in the news, Leave a comment Permalink

Apparently, in Malaysia some car thieves cut off a guy's finger to use it to start his fingerprint-protected Mercedes.

Now the only question is how the finger got to San Jose from Malaysia...


I am the NRA

posted by collin on 05.03.30 at 21:38, null, null, Leave a comment Permalink

When everyone has tiny guns there will be no crime.

The six legged majority...

posted by ben on 05.03.30 at 19:16, null, null, Leave a comment Permalink

President Hubbard's first step in establishing the RICH Economy was to offer a prize of $50,000 per year to any worker who could design a machine that would replace him or her.

When the primate labor unions raised twenty-three varieties of hell about this plan, Hubbard countered by offering $30,000 a year to all other workers replaced by such a machine. The rank-and-file union people fell into conflict immediately, some accepting this as a fine idea (this group consisting mostly of those earning less than twenty thou per annum), and the leaders still hypnotized by the conditioned and domesticated primate reflex that Employment was Good and Unemployment was Bad.

While the unions squabbled among themselves and ceased to present a united front against the RICH scenario, conservatives mounted a campaign against it on the ground that it was inflationary. Here Hubbard's political genius showed itself. She made no effort to reason with the intellectual conservatives, who were all theologians in disguise. All corporation heads and other alpha males of the right, however, were invited to a series of White House multimedia presentations on how RICH would work for them.

The chief points in these presentations were that: (1) a machine works twenty-four hours a day, not eight-thereby tripling output immediately; (2) machines do not take sick leave; (3) machines are never late for work; (4) machines do not form unions and constantly ask for higher wages and more fringe benefits; (5) machines do not take vacations; (6) machines do not harbor grudges and foul up production in sneaky, undetectable ways; (7) cybernation was advancing every decade, anyway, despite the opposition of unions, government, and these alpha males; it was better to have huge populations celebrating the reward of $30,000 to $50,000 per year for group cleverness than huge populations suffering the humility of welfare; (8) with production rising due to both cybernation and the space-cities, consumers were needed and a society on welfare was a society of very meager consumers.

The alpha males were still fighting among themselves about whether this was "sound" or not when it squeaked through Congress.

Within a year the first case of the new multi-inventive leisure class appeared. This was a Cherokee Indian named Starhawk, who had been an engine-lathe worker in Tucson. After designing himself out of that job, Starhawk had gone on to learn four other mechanical factory jobs, designed himself out of each, and now had a guaranteed income of $250,000 a year for these feats. He was now devoting himself to painting in the traditional Cherokee style-which was what he had always wanted to do, back in adolescence, before he learned that he had to work for a living.

By 1983 there were over a thousand similar cases. Many had gone on to seek advanced scientific degrees, and some had already migrated to the L5 space-cities. The swarming was beginning.

The majority of the unemployed, living comfortably on $30,000 a year, admittedly spent most of their time drinking booze, smoking weed, engaging in primate sexual acrobatics, and watching wall TV.

When moralists complained that this was a subhuman existence, Hubbard answered, "And what kind of existence did they have doing idiot jobs that machines do better?"

Some of the unemployed were beginning to seek jobs again; after all, $48,000 or $53,000 is better than $30,000. Usually, they found that higher education was required for the jobs that were still available. Many were back in college; adult education, already a fast-growth industry in the 1970s, was now the fastest growing field of all.

-Robert Anton Wilson, Shrodinger's Cat Trilogy, pg. 255-257

Also see

My evil is strong.

posted by ben on 05.03.30 at 15:11, null, null, Leave a comment Permalink
-Vermeer van Delft

HTML is not code.

posted by ben on 05.03.30 at 00:38, null, null, 14 comments Permalink

Free software can only exist where there is no real commercial competition... for example blogging software. The only real commercial offering is Movable Type and it has no multiblog support, and is written in perl... a curse significant enough to eternally damn any large piece of software.

Of course, this doesn't mean the free as in... offerings are any better. No doc, code written by 12 year olds, otaku, and unemployed devry graduates... a fundamentally flawed model for camera phones... documentation filled with ellipsis.

Open source can't work. The only good parts of Unix/Linux/whatever are left over from Bell Labs, SCO and other people who didn't suck. Remember the BSD man pages in early versions of OS X?

The most damning thing about open source is not that the code is shit or the documentation unusable... no it's the GUI, the fact that my grandmother, mother and sister could never use Linux because it's both impossible to install and to use. The problem is nerds are trying to do UI... and they're dumb. No, they don't think differently or better. They are idiots. They think people like sifting through outdated man pages and navigating flame wars on forums while waiting for the asshole who wrote their wireless driver to stop playing World of Warcraft and fix the bug that keeps borking the network stack causing kernel panic juju... which all sounds a lot like a blue screen of death to me... People don't like this, I don't like this. I like my wireless and sound to work. I like my graphics driver to support the resolution, refresh and memory my card was designed for. I like using bluetooth to connect to the internet over my phone. All these things simply work on Windows. Same with Mac OS... because they pay people to write code.

Sometime soon Microsoft or Google will enter the blog market in earnest... assuming the market is significant enough for them to care... Movable Type will disappear along with Word Press, b2evolution and even poor bloxsom. And no one will care.

comment by anwar on 05.03.30 at 06:30
ranting late at night again, huh ben?
comment by collin on 05.03.30 at 09:32
One, two, three, four.
I declare a flame war.
comment by scott on 05.03.30 at 11:06
There's no doubt that free software is currently a far cry from (some) proprietary software when it comes to out-of-the-box ease of use by non-geeks. But I think it's a mistake to assume that free software can't get better in this regard. (After all, who would have said the free software that already exists was possible before it happened?) Also, for a lot of computers, ease of installation and configuration is not as big an issue as it might seem. And finally, it is not impossible for people to get paid to write free software (witness IBM).

Free software is relatively new in the world. For one thing, this means there's just been less time to get things right. For another, the whole collaborative, over-the-Internet development approach involving lots of volunteer labor is a new thing in the world, and free software hackers are early adopters of it. In time, people with UI-related skills may get in on the game as well, and an infrastructure that allows a sufficient degree of collaboration among people with different sets of skills will emerge. There's no law that says the only people who can ever contribute to free software development have to be bad at UI design.

(I do admit that one motivator for open-source developers is being able to do things their own way, without having to deal with management and marketing and so on.)

Secondly, the ease of installing and configuring things is a big issue for home users and small businesses, but much less so for servers and even desktops in larger businesses, which have trained IT staff. I would imagine that free software is generally easier to administer on a large scale, or ultimately could be (anybody have evidence on this question?). In any case, setting up a large server installation (say) is going to require people with technical expertise, and for these people the types of usability difficulties that arise with free software aren't much of an issue.

Free software is not going away. If nothing else, there are too many people out there (like Marco, or me), who have come to positively detest using proprietary software after exposure to free software, and most of these people are among our future technical leaders. As time goes on, I think it likely that governments will increasingly require at least some software to be free when they have paid for its entire development. More and better revenue models that involve paying people to develop free software will be created.

What I imagine for the future is an ever-growing substrate of free software, and more proprietary software being written to be used with free software, either made-to-order for a particular client, or in products like OS X. Software or services that ease the use of systems consisting largely or entirely of free software will emerge for various business sectors.

In the end, I think there's no question that a given piece of software under a free (or open source) license is more valuable than the same piece of software under a proprietary license. Therefore, in some sense, producing free software is a Good Thing. The question is whether we can come up with ways to direct enough of that value to the people producing the software as compensation. I think that the added value of free software over proprietary software is high enough for enough people that we can probably figure out a way to compensate its developers.

The win for free software is biggest for software that lots of other software uses or builds on, like operating systems and databases. In the long run I think this sort of core system software will be all or mostly free software, and anyone in any sector of the proprietary software business who neglects to think seriously about how free software relates to their future does so at their peril.
comment by ben on 05.03.30 at 12:38

"In the end, I think there's no question that a given piece of software under a free (or open source) license is more valuable than the same piece of software under a proprietary license."

One of the best things about the Microsoft APIs is that the doc is very good. If this went free the doc would almost immediately go to shit. Script kiddies would post their crap hacks to the OS. They would post their crap hacks to the doc. Without a unified despotic vision (we love you Bill) the software would become unusable and the doc would end up the Swiss cheese on man pages, tech forums and blogs that is Linux doc.

I have no doubt that people like you and even me will continue to use open source because we can hack it to do weird things... even if the support is made up of morons. But, I'm not sure the commercial software that most people use can ever can be free... precisely because of its ease of use. IBM pays people to code free software, but makes money back charging for tech support. If the software actually installed/maintained itself that wouldn't be an issue and IBM would need yet another business model.

I firmly believe that, more than any smaller order open source trends, the software industry is moving toward automation. Higher levels of abstraction, self configuring programs, self patching programs, programs with reactive network security models... these are all things that are coming... and I don't see them coming from the free community but from the twin gods of Google and Microsoft (well maybe the Mac at 3 - 6% too).

Mostly I'm angry about the stupid things they did in b2evolution. Though now that it's morning (well, afternoon) things seem both better and fixed.

comment by It is me on 05.03.30 at 14:33
Is this why you steal the credits from the makers of this b2Evolution (free open source)
blog system, that is this, you use?
Are you ashame everyboby will know that you shout loud and then hide the b2Evolution logo?
comment by devin on 05.03.30 at 19:34
Hmmm . . . Asbestos suits required.

Ben, your arguments might be more persuasive if you knew what you were talking about.

You write:

Free software can only exist where there is no real commercial competition... for example blogging software. The only real commercial offering is Movable Type and it has no multiblog support, and is written in perl... a curse significant enough to eternally damn any large piece of software.
. . .
Sometime soon Microsoft or Google will enter the blog market in earnest... assuming the market is significant enough for them to care... Movable Type will disappear along with Word Press, b2evolution and even poor bloxsom. And no one will care.

Google of course, does have blogging software. They have hundreds of thousands of users. Whether they have entered the blog market "in earnest" is, I suppose, subject to debate.

Microsoft also has blogging software, too, although they don't seem to quite know what to do with it.

Yahoo just announced a beta blog program, too.

These are all reasonably well designed, have well-tested interfaces, are reasonably full featured — but less so than Movable Type. If they are successful, though, it will probably be because they are free.

You write that MT is doomed to fail because it is written in perl — in fact, if Movable Type succeeds it will be precisely because it is written in perl. Movable type exposes its guts in a well-documented, well-designed way. These means that it is customizable and that you can share your customizations with others.

You also write:

Open source can't work. The only good parts of Unix/Linux/whatever are left over from Bell Labs, SCO and other people who didn't suck. Remember the BSD man pages in early versions of OS X?

But consider, for example, all the things that Berkeley and BSD have given us:
  • TCP/IP — Although TCP/IP was developed at DARPA, it only came into wide use after 4BSD implemented it. In fact, BSD's code was used as the basis for almost all TCP/IP stacks — from Solaris to Microsoft Windows.
  • sockets — 4BSD also gave us sockets, a sort of virtualized internet connection that revolutionized inter-process communication. AT&T/Bell Labs/USL created a rival API, called streams, which wasn't designed to multiplex. sockets eventually won, and is the dominant metaphor for internet connections.
  • tty — BSD gave us the tty system for terminal independent screen support. Ben, this made your beloved vi possible. AT&T et. al supported the rival termio.
  • Demand paging — BSD was the first UNIX to implement demand paging, in which parts of a program are only loaded if needed.
  • File systems — For a long time BSD was the source the most exciting filesystem developments: they gave us FFS, the first Unix file system that wasn't awful, as well as Sprite FS, the first logging file system.

Ben, most of the things we think about when we think about Unix were in fact developed outside AT&T. It was only the open source nature of the Unix development process that allowed these innovations to quickly move between academic, corporate, and hobbyist environments.

I'm surprised that you mention MacOS X's out-of-sync man pages; it seems like a good counter-example to your argument. After all, Apple paid someone to sync their BSD tools and they messed up.

You write:

The most damning thing about open source is not that the code is shit or the documentation unusable... no it's the GUI, the fact that my grandmother, mother and sister could never use Linux because it's both impossible to install and to use. The problem is nerds are trying to do UI... and they're dumb. No, they don't think differently or better. They are idiots.

I think you are correct that bad user interfaces are the biggest drawback to most open source programs. In fact, the only open source program I can think of with what I would consider a good UI is FireFox. But this is because UI is fundamentally hard. Everybody gets it wrong. There are probably only ten or fifteen people in the whole world qualified to design user interfaces. Luckily, they can release guidelines, design tools, APIs, and libraries to make it easier for the rest of us. The Unix world hasn't, until recently at least, had these people working for them.

In comparison, I think, open source excels at programs used for communicating, while corporate-backed ones tend to be awful. It's no coincidence that Apache is so widely used and that Microsoft's IIS is so bad. Similarly with OpenSSH (now) and the commercial SSHs. The best XML parsers are open source, as is almost any new programming language interpreter/compiler. In situations where you want transparency and interoperability, open source is king. It's not hard to see why,

In a networked world, interoperability is usually more important that UI. This is why Windows 3.1 beat the vastly superior Macintosh operating system. It's why we still use QWERTY keyboards. It's why webmail beats Outlook, even though webmail has a uniformly awful interface.

You say most open source software sucks, most open source developers are idiots. If this is true, it is only because most software sucks and most people are idiots.

As with everything, you have to make decisions about tradeoffs (duh) among usability, cost, performance, maintenance, and control.

But, of course, you can't make these decisions if you don't know what you're talking about.
comment by ben on 05.03.30 at 20:07

...Dug hole with late night rant... continued with lunch... will continue until the end of time...

The majority of bloggers don't use full featured software. They use things like Blogger and Live Journal. They may be happy with these, they may not. They will never know because they can't use software like Movable Type, Word Press or b2evolution. It's too difficult to install, maintain or even put up a picture. I fully accept there may not be a market for a full featured blogging suite like Movable Type to replace Blogger. What I do not accept is that people do not want the features real blogging software has... just because they don't know of/can't figure out how to use these features.

If I were Google or Microsoft (I know), I would set my vast teams of software engineers, replete with actual degrees, on adding those features in a way people can use... a way that involves no command line, no ftp and no source. I assume Microsoft has some plan along those lines. Of course, then there's the question of how to make money on any of this...

As for the rest, I think you are arguing that open source makes for good standards. This is true of your examples and I largely agree with you, but then there's the W3C... I like my CSS and HTML to behave the way I expect it to, not the "right" way. Open source may even be good for developing backends... Apache is nice. But, as I said at lunch, Apache may only appear to be good. What if there were a web server without httpd.conf, .htaccess and apachectl? Something my little cousin could use... Then webservers might not be run just by sysadmins, but by everyone. Webservers might be used to do new things.

I can envision a scenario where everyone has their own server on their stylish mac laptop and they browse idly around the adhoc net at the coffee shop... seeing what their neighbors are up to.

I think usability of software will be the most important thing in coming years. We are getting more education; our time is hence becoming more valuable, so there's a higher cost to wasting time on hard to use software. In the next few years I think people will be willing to pay ever higher premiums for intuitive software. The open source model is dependent on people who enjoy writing code. The problem is these people write code assuming that other people think like them and want to see code on a regular basis. Right now the most popular way to avoid seeing code is a Microsoft GUI... maybe it will be something better soon... but I don't imagine that something will be the command line.

comment by scott on 05.03.31 at 01:17
Well, Ben sure is getting the hammer put down on him for one frustrated rant--good work, boys ;).

I absolutely agree that UI is the major (or at least a major) factor in the future of computing. But what does this mean for free software?

The reason hackers like to see code (and be able to change it) is that it allows them to make computers do what they want them to do. The problem of how to give people access to all the functionality a highly skilled user can get by hacking around b2evo or Movable Type, and making it as easy to access this functionality as it is to use Blogger, is very difficult and not yet solved.

What would it mean to really solve that problem? At an extreme, it would mean that the average non-geek user can get all the functionality and customization out of computers that skilled hackers can currently get by tweaking and programming. In a sense, it means everyone being able to program computers, where programming is something much different and much easier than what that word means today.

Now, if this is possible, what does it mean for open source? People doing the future, easy equivalent to programming will probably be very happy to share around the fruits of their labor without compensation. So their "code" is likely to be open source.

The engine(s) that make this easier "programming" possible, which will probably be created using something more similar to current programming techniques and will require professional-level skills to build and modify will be in a space where transparency and interoperability matter *a lot*, much like networking protocols and compilers and other areas where open source dominates today. So they, too, stand a pretty good change of being free software.
comment by scott on 05.03.31 at 01:24
Another tidbit to consider is that people who really don't know how to use computers, e.g., non-technical people who didn't grow up with them, have probably been put into the fuming state in which Ben wrote the original post by nearly every piece of software they have ever used, proprietary or otherwise. So proprietary software doesn't have especially much to brag about in the UI department, either.

I think this also points out that some amount of education is always going to be required, although it may become small enough that people absorb it more or less naturally in the course of growing up.
comment by marco on 05.03.31 at 09:37
Okay, my turn to chime in here.

First, I agree with all that has been said about the UI of free software generally being not up-to-par. Much of this comes from the fact that the people writing free software tend to write it as they would like to use it, which means it's written for your average hacker-programmer rather than your average user, and it often has personal quirks or strange preferences in evidence.

There's reason to believe this is changing, though, at least in some areas. A significant portion of the code-writing community is currently engaged in actively working on making GUIs easy to use for everyone. The brunt of this work is going on in the Gnome and KDE projects, and if you take a complete Gnome or KDE installation, these days it's pretty close to Windows or Mac OS as far as finding programs to run in menus, navigating the filesystem, and interacting with programs. I can't say much about this as I don't use many of those features--I personally like my command line and eschew the silly desktop icons and graphical file browsers Gnome tries to get me to use. =) But it seems to me that as far as basic user interaction with the operating system, it's getting pretty close to where MS and Apple are.

Of course, that's not the whole story: there are also the applications themselves, as well as all the utilities (file search, etc.) and everything. Ben, I agree with you--I'd love for apache to be easy enough to administer that anyone could easily run it on their laptop, showing whatever crazy stuff they want their fellow coffee drinkers to know about them. Part of what it'll take to get there, though, isn't so much the ability for users to configure apache in an accessible way as intelligence in setting defaults and automatically configuring it the right way--for example, enabling "coffeshop mode" with one toggle switch. MS is probably moving towards that kind of thing faster than the free/open-source community, but I don't see any reason why it should be inherently easier for them.

Some linux distros are also closer to the "it just works" end of things...I've heard Mandrake is pretty easy to install and has drivers for just about everything, and Redhat has seemed to be going that way too. Again, I don't know about this from personal experience. ;)

But I also want to say a couple of things about the usability of proprietary software. As Scott pointed out, many people have plenty of problems using Windows or Mac OS. It's also not as if MS provides everything users want in an interface--they haven't updated IE in how long now? It still lacks popup blocking and tabbing (or it did a few months or a year ago, whenever the last time was that I had any interaction with it), and MS hasn't exactly been keen to get on this despite popup windows being a major annoyance for most people. And the auto-format stuff in Word--aside from spelling correction and a couple other ones like that, most of it is very annoying...I've seen people try frustratingly to figure out how to turn it off much more often than actually use it. And when I glance at the tech advice column in the Chronicle, more often than not there's a letter of the form "how do I turn off X stupid feature in Word?" And I think I know of one person who uses Clippy and many more who wish he would die a painful death.

Okay, enough on that for now.
comment by ben on 05.03.31 at 13:19

In favor on IE...

It does block popups and has for some time now. There are no tabs, but tabs seem to be against the IE ethos. The argument I've heard for why we haven't seen an entirely new version of IE is that Microsoft has decided the browser is dead. Apparently something strange is going to happen with Longhorn.

I already like that my web browser is the file system browser. I also like how fast new IE windows load. And the typical end user cares even less about the morality of those fast loading windows than I do.

I went to the IBM and Microsoft jobs sites in Firefox yesterday (I've been using it since I start the work on this site) and had weird problems searching for jobs. The thing I like most about IE is that it just works. I've never had to browse to something in IE, realize it isn't going to work and then use Firefox / Mozilla / Netscape / Safari instead. I've have had the opposite problem with all those browsers.

IE has a 90 some % installed base. Whatever you think of how they got it, there are benefits to being part of it.

comment by ben on 05.03.31 at 13:32

As far as the command line...

Marco, you say that you prefer the command line to the GUI:

I can't say much about this as I don't use many of those features--I personally like my command line and eschew the silly desktop icons and graphical file browsers Gnome tries to get me to use. =)

I suspect that if you were presented with a more usable GUI, you would find the opposite. The best example I can think of is Mac OS.

Devin is certainly no less proficient in the use of UNIX than any of the rest of us, yet he uses the Mac GUI all the time instead of terminal windows. Admitedly he uses terminal windows as well, but he does use the actually GUI. On the other hand, when I use a Linux box, I use terminal windows and some web browsers. I suspect the rest of you do much the same. I suspect a large portion of the Linux population use the Linux GUI for managing terminal windows and little else. I see that as a sign of the failure of the GUI and not an innate preference for numerous terminal windows.

comment by ben on 05.03.31 at 13:44

Marco, you say the way to fix apache to do what I suggested is:

for example, enabling "coffeshop mode" with one toggle switch

I would argue this philosophy is why GUIs are so unusable. I don't want vast arrays of switches for multiple modes. I want the thing to work on its own. If it fails to work on its own, then I'll take a look at the GUI which should tell me what it thinks is wrong and how I might remedy it (but I only want to see this if the computer has already tried to fix the problem itself). If all that fails, as it often does, then I end up fixing something on the command line. I want that final necessity to go away.

comment by marco on 05.03.31 at 14:32
Ben, I'm not sure we're talking about the same aspects of the "coffeshop mode" thing. What I meant was that this mode (like many other configuration choices) is not something everyone will want automatically on all the time, so there will have to be some convenient way to say when it is on and when it is off. One way would be to have some little switch to turn it on when you go into the coffee shop and off when you go into a job interview, or whatever. Sure, it'd be nice if the thing could "work on its own" and read your mind about when you want it on and when you want it off, but realistically you're going to have to tell it somehow. Now if the average user is going to be able to do this, you're not going to have the luxury of letting the user decide, for example, how many connections to allow at once or whether to use SSL. If it's going to be usable by everyone in that coffeeshop, it's going to have to be on the level of "do you want to turn this feature on?" Much more than that and you're going to have to ask for full natural language understanding to configure it.

About the command line: I suspect Devin's preferences are just different from mine. There are several reasons why I like using the command line, and I think it'll be a while before any GUI offers all these things (but when it does, sign me up!).

One thing I like is that that finding things through the command line requires only typing, not physical movement/accuracy with the mouse. I find it tiresome to have to move the mouse accurately onto a small icon, pause the movement, double-click, move to the next icon, pause, double-click, and so on just to open up a file that's a few levels deep in my directory tree. Sure, if I used a GUI interface more I'd have shortcuts set up, and I'd have a window open to my current working directory and all that, but if I have to open up a file in another directory I'd rather type than mouse-navigate to find it. Part of this is the fact that tab completion makes it so easy (this is far and away my most-used feature of my computer)--it takes so few keystrokes to get to where I want.

Another reason why I like the command line is the ability to tie commands together with pipelining or scripting. For example, one thing I do pretty frequently is search with 'find documents/ |grep "somestring"' or the like, to find documents (or emails or whatever) that have some particular text in them. Sure there are find features in Windows too, but they require clicking on the menu and the Find icon (or maybe pressing Menu+F or something), then browsing to the directory you want with the mouse (ugh!), then clicking in the text field (another stupid mouse click!), maybe checking or unchecking some boxes for options, then finally typing the string you want to look for. No thanks. If it'll work so I can just type 'Menu+F documents/ TAB somestring ENTER' with appropriate focus in the windows popping up, then that'll be pretty close to what I'm used to, but those things tend to be designed with the mouse at the center of everything.

On the scripting side, I've recently discovered the joys of 'perl -e -p "s/out/in/g" < infile > outfile' for string substitutions in files. No more opening of files in an editor to do substitutions!

Another aspect of it is that I don't like moving my hands between the keyboard and mouse so much--if my hands are already on the keyboard, I'd much rather keep them there to open files, access features, or whatever, rather than moving over to the mouse. On that note, I actually like the little rubber joystick mouse on my ThinkPad quite a bit, and that takes away part of this reason to favor the keyboard.

There are more reasons, but I have to get going to class pretty soon. It mostly boils down to the fact that physical navigation with the mouse seems clunky to me compared with logical navigation through the keyboard. And like I said, it's really just a personal preference--I'm certainly not claiming more people should use the command line instead of a GUI. And when a GUI offers me all the conveniences I've come to expect from the terminal, I'll gladly switch to it.


For those of us lacking fine motor control...

posted by anwar on 05.03.28 at 12:08, Do It Yourself, null, 1 comment Permalink

Now we can use surface mount (and otherwise hard to solder) parts in our prototype boards...Yea!

Soldering with your toaster-oven

Design your own PCBs

Note to self: buy a *spare* toaster oven.

comment by collin on 05.03.28 at 12:41

Though their CAD software saves in a proprietary format, boooo.

the transparent economy

posted by scott on 05.03.28 at 00:10, Catch-all, null, 4 comments Permalink

I've been thinking for a while that it'd be really desirable when purchasing a product to be able to learn various things about how that product was made. As it stands, if Company A figures out a way to make a widget for 20% cheaper than Company B, they can pass that savings on to the customer and sell a heck of a lot more widgets than B. If B doesn't do something to catch up, they're in deep trouble.

On the other hand, if Company A makes its widgets for the same cost and sells them for the same price, but causes 20% less environmental harm per widget (or gives its employees 20% better health insurance), it's possible no one will even know the difference.

It would be a Really Good Thing if the same competitive pressures that drive down costs could be made to drive down waste and pollution, or to drive up the quality of life of various workers. One way to get at this is Pigovian taxes, which cause information about the impact of making something to be reflected in its price (e.g. by charging a higher tax for products that required more pollution to produce). It seems like another way would be to make more information available to consumers. (Collecting such information would also provide a good basis for imposing Pigovian taxes.)

It seems like we're nearing the point, technologically, where a supply chain could be designed that would keep track of just about every detail regarding a product's manufacture and shipping, and make this information available (e.g. by scanning a barcode). As far as I'm concerned it is very much my business to know all sorts of details about how the things I'm thinking of buying were made.

Any thoughts?

comment by graham on 05.03.28 at 01:43
This has been a fairly hot topic in the Seattle area.

The technology for tracking a product's manufacture and shipping already exists in the form of RFid tags with built-in sensors. For example, I've seen tags used to track the temperature of produce during shipping in order to determine the remaining shelf life of the items.

From my two experiences with manufacturing, however, I think it might be difficult to put a number on how much pollution is generated in the manufacture of a product. Instead (and this is what most of what I have read on this subject says) it might be more practical to tax products based on it's materials and how easy it is to reclaim them. I've heard of several companies starting to design their electronics products to facilitate dissasembly in order to remove the PCBs for proper disposal and recycle the plastic housings.

[argh... stupid tiny comment box]
comment by anwar on 05.03.28 at 11:40
So, I was thinking about this idea a while back too. Originally I believed that the best place to start was by creating graphs of all the components that went into a product. However, for any reasonably complex thing (e.g. a computer) this graph quickly spirals out of control.

Now I tend to agree with Graham about the materials/energy idea. On the first order, if you know what the materials used in the product are (and also the energy/pollution required to mine/extract/synthesize them). You could quickly come up with an envelope for its total environmental cost.

More on this later...
comment by marco on 05.03.28 at 12:28
At a presentation about food production a year or two ago I remember hearing about some northern European country (Denmark?) that has an interesting system already in place. Meat sold in stores has two bar codes: one is for the regular check-out and price/inventory information, but the other can be scanned at stations in the store to tell you about the production history of that particular item. It sounds like a really cool step in the direction you're talking about, Scott, though I don't know any details about it. Might be worth checking up on.
comment by devin on 05.03.28 at 15:25
One of my college roommates (Chris, for those of you who have meet him) worked at an institute in Germany one summer where they tried to calculate the total amount of various types of resources needed to make consumer goods. Under this kind of system, computers (of any kind) are really expensive. I seem to recall that the total environmental cost of a microwave was absurdly high compared to that of a regular convection oven.


Cricket Guerillas

posted by collin on 05.03.27 at 18:26, null, null, Leave a comment Permalink

Of course crickets can launch missles.

Party like it's 1996.

posted by ben on 05.03.27 at 14:51, null, null, Leave a comment Permalink

Yahoo bought Flickr. And that's all well and good, but the model seems moronic. From an Ars story:

"...Yahoo's recent acquisition of Flickr, a community-based photo sharing and indexing site, represents something of a return to the company's roots. Flickr hosts user-submitted photos, which are then tagged with descriptive metadata by community members..."

So, let me get this straight... Over the last 10 years Yahoo has lost market share to Google because having humans index web pages is too time intensive. Finally, after going from the 1st most used search engine to the 3rd, Yahoo acquiesces and switches over to machine indexing. And now they're going to repeat the process with pictures? Are they trying to run themselves into the ground? Isn't it obvious that very soon machines, not people, are going to index pictures?

Sappho's Leap

posted by ben on 05.03.27 at 14:29, null, null, Leave a comment Permalink

In my dreams I paint
Like Vermeer van Delft
I speak fluent Greek
And not only with the living.

-Wislawa Szymborska
"In Praise of Dreams"


posted by collin on 05.03.26 at 18:11, null, null, Leave a comment Permalink

Wasted Years
John Lee Hooker

Wasted years been brainwashed by lies
Oh yes I have
Oh wasted years
I'm talking about wasted years
Oh I'm not seeing eye-to-eye
I just can't see the things I should see
Wasted years, baby
I was taking the wrong advice
I know you was, I know you was
And I was too
All alone I'm travelling
Travelling through these wasted years
For so long, so long, so long I was
Oh, I must have gained some wisdom
Down through the years I did
Somewhere along the way
Oh yes, I did, oh yes I did
Thet's why there can't be no more
No more
No more wasted years today
I got wise, I got wise to myself
Well baby the great sadness
Oh, you've got to let it all go
Oh yeah, oh yeah Van
Live in the present
Live in the future John Lee, ain't that so
Oh, it's a sad feeling, oh yeah
Oh, you've gotta find something
To carry you through, carry you through,
carry you through
I've learned my lesson
I ain't gonna do it no more, yeah
Now Van
Now John
I've learned my lesson
I should have a long time ago
That's right
All these wasted years, wasted years
I finally woke up and got wise
I ain't gonna be, ain't gonna be no fool no more
Now Van, now Van
Ain't gonna be nobody's body's fool no more
Sing the song Van, sing it with me
Well all alone, all alone I've been travelling
Travelling all along through these wasted years
Dark, dark wasted years
So dark here
Dark, dark, dark, dark wasted years
I must have gained something
Oh travelling along the lonely way
Yeah, I've learned a lesson
I'm gonna make damn sure baby, make damn sure
There's no more wasted years today

We all need more ocenaographic data.

posted by collin on 05.03.26 at 17:55, null, null, 1 comment Permalink

Ride the sea glider to Bermuda!

comment by graham on 05.03.26 at 22:52
Jeez, $3000 for batteries?
They should use stirling engines instead. That would be super-cool.

posted by ben on 05.03.26 at 02:02, null, null, Leave a comment Permalink

Give me good without evil... fundamentalism giving way to kindness all under the banner of ferocious humanism.

Give me heroism without slaughter. Life without sadness, love without end. Give me lust without hurt. Indecency without horror... Art without failure.

Give us a world in which banality falls to thought, where truth and beauty are not abstractions and metaphor is not hyperbole... where age implies only nobility and wisdom... and wisdom is wealth.

Give us a world where all these things are true and eternal... knowable and unalienable... a world where I place freedom, liberty and justice on the coffee table next to a book on Frank Gehry that never needs dusted off.

A Fairly Honourable Defeat

posted by ben on 05.03.26 at 00:57, null, null, Leave a comment Permalink

Brock exhibited this life-size female nude as a plaster cast at the Royal Academy in 1898. He completed the marble in the following year, and showed it at the Paris Universal Exhibition. Critics praised its combination of naturalism and spiritualism, as well its subtlety in modelling and expression of feeling.

Unusually, Eve is not presented as a temptress. Instead she is shown as self-absorbed, her head bowed as if in shame and her left arm placed protectively across her chest. Brock also produced a number of smaller bronze replicas of this subject.

-Tate Britain caption

We need great golden copulations.

posted by ben on 05.03.26 at 00:38, null, null, Leave a comment Permalink

Art doesn't have to end happily, it only has to end beautifully. Julie in Blue having sex with a man she doesn't love... only to feel. Or the dying baron flung by a branch to the sea... this is art... Art that ends in death, pain and beauty... a woman in a bath tub watching the man she doesn't love. Or the wet body of the girl who played go. All these things are sad, beautiful and somehow true... maybe even more true than the ugly.

How things ought to be:

Truth implies Beauty implies Platonic Essence.
Beauty does not imply Truth.
Ugly implies Falsehood does not imply Nonexistence.
Ugly implies Nonexistence.

Ergo Juliette Binoche is Beautiful therefore has Platonic Essence, but is not necessarily True.

How I hope. I should install the LaTeX package so we can do some formal aesthetic logic... philosophy is so many lies. I saw Ted tonight. His arms are still amusing.


posted by graham on 05.03.25 at 15:02, Photos, null, 2 comments Permalink
comment by collin on 05.03.25 at 17:05
Pingpong balls in an orange lit, neoclassical, train station?
comment by graham on 05.03.25 at 19:17
Good guess... it's a University student art installation in a roman-era cistern under Istanbul.

the thing with the stuff

posted by collin on 05.03.25 at 14:03, math, null, Leave a comment Permalink

Can you put bounds on the number of self-assembling DNA structure of a given type by formalising to a commutative diagram? Can you look at the dynamics generated by that diagram and show there exists a bijection between cycles on disjoint subsets of the space to set of well formed structures?

I have no fucking clue.



posted by marco on 05.03.24 at 17:43, funny, null, 1 comment Permalink

comment by Scott on 05.03.27 at 23:46
This rules!!


Failure-Oblivious Computing

posted by devin on 05.03.23 at 23:05, Technical, null, Leave a comment Permalink

This is a fascinating paper on Failure-Oblivious Computing, in which bad memory writes are ignored and bad reads return manufactured results. The authors contend that in servers this increases availability and (!) security.

I'm not sure what I think about this, yet, but it is an interesting read.

Karakoy Warehouse

posted by graham on 05.03.23 at 21:00, Photos, null, Leave a comment Permalink

On Liberty

posted by ben on 05.03.23 at 14:34, null, null, 1 comment Permalink

If all mankind minus one were of one opinion, and only one person were of the contrary opinion, mankind would be no more justified in silencing that one person, than he, if he had the power, would be justified in silencing mankind.

-John Stuart Mill, On Liberty, page 18-19

Genius can only breathe freely in an atmosphere of freedom… If from timidity they consent to be forced into one of these moulds, and to let all that part of themselves which cannot expand under the pressure remain unexpanded, society will be little better for their genius.

-John Stuart Mill, On Liberty, page 62-63

What has made the European family of nations an improving, instead of stationary portion of mankind? Not any superior excellence in them, which, when it exists, exists as the effect not as the cause; but their remarkable diversity of character and culture.

-John Stuart Mill, On Liberty, page 70

jsm2 jsm3

Whenever, in short, there is a definite damage, or a definite risk of damage, either to an individual or to the public, the case is taken out of the province of liberty, and placed in that of morality of law.

-John Stuart Mill, On Liberty, page 78-79

Is it not almost a self-evident axiom, that the State should require and compel the education, up to a certain standard, of every human being who is born its citizen?

-John Stuart Mill, On Liberty, page 100

It still remains unrecognized, that to bring a child into existence without a fair prospect of being able, not only to provide food for its body, but instruction and training for its mind, is a moral crime, both against the unfortunate offspring and against society.

-John Stuart Mill, On Liberty, page 101
comment by Other Graham on 05.03.23 at 22:14

If all mankind minus one were of one opinion, and only one person were of the contrary opinion, mankind would be no more justified in silencing that one person, than he, if he had the power, would be justified in silencing mankind.

This reminds me of a hung jury. The other members may not be allowed to silence the dissenter, but they can badger the poor soul almost indefinitely.

The Henry Miller Award

posted by ben on 05.03.23 at 13:38, null, null, Leave a comment Permalink

"Say it," he said.
"I am a slut," she said.
"What else?" he said. She looked at him, thinking to himself. She brought the pen up to her breasts and wrote Bitch between her nipples. "Good. Good. I'll get the mirror," he said.
He grabbed a large mirror from the bathroom and brought it out to where his wife lay. He set it against a foot stool. He grabbed a blue pen and began marking on her body himself. He drew a pair of arrows pointing to her vagina. He gave her a fake black eye. She gave up her skin for him. He marked her carelessly, waiting for something to appear in the blemishes. Then she watched her body in the mirror as he entered her from behind. He watched too, and they became angry with excitement. She whimpered a little as he moved. "You fuck like a dog. You fuck like a dirty dog," she told him.
"What's your pussy feel like? What's my dick feel like?"
"My pussy is wet. Your cock is hard."
"Is this too hard for ya', grandma?"
"No way, old man. You got to tear my pussy up."
A puff of air blew out of Joseph's mouth like a tire exploding. His penis was out and it oozed a small amount of semen. Helen flattened underneath him, quivering slightly herself. Both bodies sucking air. Helen twisted around and smeared his goo across her belly, smudging the words there. Joseph spooned himself against her. "That was a good one," Helen panted. Joseph chuckled lightly into her back. "What's so funny?" she asked.
Joseph rubbed some of the black marker from his eyes onto Helen's shoulder blades. "Bang me," he blurted, then started laughing more freely.
Helen turned around, put her hands around his throat, an imaginary choke hold. "Well, we're almost to that point, aren't we? You better watch out next time or you might be the slut." They held each other for a moment and felt their breathing synchronize. "Your birthday's coming up. I'll make sure to put a dildo on the list," she told him. "For the man who has everything."
Between the cracks in the blinds, they saw the early night darken. Each of them could swear that they could hear the other's heart beating.

-from Kevin Sampsell, Beautiful Blemish
Stolen from

I liked the Indigo Girls before I started dating lesbians.

posted by ben on 05.03.23 at 13:30, null, null, Leave a comment Permalink

At night I cry. He holds me and I think of her. He says it will be ok. Then I fall asleep. I love her.

In the morning the sun rises and I think of her. She’s still asleep. Her long brown hair spreads out over the pillow. A leg hangs carelessly off the bed, as the sun creeps up to her left eye and she sneezes. Still asleep, she smiles slightly and rolls over. Her back is freckled and tan from this morning routine. She sleeps naked.

I sleep naked too. It’s more natural, healthier I’ve heard. Feeling the sheets on my bare legs gives me a hard on. I reach down to check if it’s really hard. It is. This means nothing. In the morning the sheets are stained and crusty in one spot. I take the sheets off and, wearing them as a toga, walk to Laundromat.


The Gatekeeper Model

posted by devin on 05.03.22 at 14:02, Interesting, null, Leave a comment Permalink

There is a great excerpt from an interview John Batelle (BoingBoing!'s web/ad guru and search engine watcher) did with Barry Diller (former CEO of Paramount and Fox; his company has just bought Ask Jeeves):

JB: For now, there are still vast differences [between traditional media and the internet]: Old media tends to be closed; the Web is open. Will the closed model stand?

BD:The gatekeeper model? No, it can't. How can you trust a gatekeeper when you know the purpose of a gate is to open and close? . . .

Barry Diller has a bit part in a book I've been reading, Disneywar, about Michael Eisner's rise and fall at Disney. It's been fascinating to to read about the fallout when "new" and "old" media combined (AOL and Time-Warner, Disney's disastrous Diller sounds like he gets it (at least as much as anyone really does).

thinking about search engines

posted by ben on 05.03.22 at 11:59, null, null, Leave a comment Permalink

A simpler version of the Cluster-Kernel SVM algorithm I came up with for my thesis might be good for search engines. It should perform about as well NN given my previous results. But, with a search engine evaluation time is probably a significant issue. So, evaluating k rather than n points ought to be a good thing.

(1) Cluster data using k-means.
(2) Compute k centroids.
(3) Normalize centroid distances.
(1) Correct for normalization.
(2) Compute nearest centroid.

This generalizes to multi-category problems very easily. If you want c categories, choose the c nearest centroids. Or, to give an undetermined number of classes, the centroids closed then some margin could be chosen.

OMG, it's full of stars!

posted by graham on 05.03.22 at 05:09, Misc, null, Leave a comment Permalink

This is a single image of a patch of space one tenth the size of the full moon, recorded by the Hubble space telescope from September 24, 2003 through January 16, 2004, a total of 1 million seconds spread out over 800 exposures. It depicts an estimated 10,000 galaxies, some up to 13 billion years old. For reference, the entire universe is 13.7 billion years old. This image was recorded in visible light.



Man Finds Satan in Turtle

posted by devin on 05.03.21 at 21:57, Absurdities, null, 2 comments Permalink

And it makes

Lucky the Satan Turtle
comment by marco on 05.03.22 at 10:35
Satan my ass. The face on the turtle is quite obviously that of Bowser, King of the Koopas.
comment by marco on 05.03.22 at 10:43
And you can also bid for his amazing story on eBay, starting bid only $499 for one DVD.

things I need...

posted by collin on 05.03.21 at 14:25, null, null, 1 comment Permalink

iron filings
heavy weight oil
car batteries

Oh, wait I need to work. More on this later...

comment by graham on 05.03.25 at 19:23
is this for that torus of death? er no, that was mercury.

what is this for?

The Bard of Nowhere

posted by ben on 05.03.21 at 12:57, null, null, Leave a comment Permalink

At Burnmouth the door hangs from a broken hinge
And the fire is out.

The windows of Shore empty sockets
And the hearth coldness.

At Bunertoon the small drains are choked.
Thrushes sing in the chimney.

Stars shine through the roofbeams of Scar.
No flame is needed
To warn ghost and nettle and rat.

Greenhill is sunk in a new bog.
No bending woman
Blows russet wind through squares of ancient turf.

The Moss is a tumble of stones.
That one black stone
Is the stone where the hearth fire was rooted.

-George Mackay, from "Dead Fires"

IBM warranty service is amazing

posted by marco on 05.03.21 at 11:24, computers, null, Leave a comment Permalink

A week ago, my laptop's hard drive croaked. Ironically, it happened while I was in the middle of backing up my data from it to my new desktop. I didn't get all the data copied, but I did get most of it. That was the first time I've backed up the data on that computer in a year or two, so it would be fair to say I'm a lucky bastard.

Today I finally got around to submitting a warranty service request to IBM (it's under a 3-year warranty that expires this June). I had to register with their service website, and then I submitted the service request at 9:35 am.

At 10:10 am, I received a call. It went something like this:

IBM: "Hi, I'm so-and-so with the IBM repair center. Is this Marco?"

Me: "Yes, this is Marco."

IBM: "I have a repair ticket here for a laptop hard drive. What is the FRU number on the drive?"

Me: "Sorry, I don't have the computer in front of me."

IBM: "What's the size of the drive?"

Me: "20 GB."

IBM: "Okay, I'll have a replacement drive sent out to you. It should be there in two business days."

Me: "Wow, that was easy. Thank you."

Amazing. All tech support should be that easy.


A courtyard at night...

posted by collin on 05.03.19 at 00:12, pictures, null, Leave a comment Permalink


I hate LaTeX

posted by collin on 05.03.18 at 20:00, math, null, Leave a comment Permalink

Render this to see the braid word for the title graphic...

$\sigma_{1} \sigma_{3} \sigma_{1} \sigma_{4}^{-1} \sigma_{2} \sigma_{4}^{-1} \sigma_{2} \sigma_{4}^{-1} \sigma_{3} \sigma_{2}^{-1} \sigma_{4}^{-1} \sigma_{1} \sigma_{3} \sigma_{1} \sigma_{4}^{-1} \sigma_{2} \sigma_{4}^{-1} \sigma_{2} \sigma_{4}^{-1} \sigma_{3} \sigma_{2}^{-1} \sigma_{4}^{-1} \sigma_{1}^{-1} \sigma_{3}^{-1} \sigma_{2} \sigma_{1}^{-1} \sigma_{3} \sigma_{1}^{-1} \sigma_{4} \sigma_{3} \sigma_{4} \sigma_{2} \sigma_{4}$

Oh, and it's pronounced like the rubber, fuckers.

Beloved Blue Bic

posted by ben on 05.03.18 at 13:23, null, null, 1 comment Permalink

I am entranced by this blogging software. It's amazing how easy it is to customize. Right now Collin is sitting across from me almost certainly breaking his style sheet and I'm doing the same and the results are completely isolated.

I don't understand how someone could start using this software naively though. I have a fair knowledge of HTML, a passing knowledge of CSS and I know how to program, so PHP isn't that strange. This certainly isn't the sort of thing my grandmother could pick up, but for open source it's surprisingly usable. And, there's support. Overall I am very happy.

The main problem is how images are handled. They are all uploaded to a single directory. There are almost certainly going to be name conflicts and the number of files in the directory will grow unmanageable quickly. There also doesn't seem to be a way to upload multiple images. However, there should be plugins to fix all these issues

Of course, I still have to figure out how to import the old wasabi and put the archive links and search links back on my page.

comment by collin on 05.03.18 at 22:20
That's a red pen, not a blue pen.


First post

posted by marco on 05.03.17 at 17:17, misc, null, Leave a comment Permalink

Just a quick first post to try this out...thanks, Ben, for setting it up.

Here we go...

posted by scott on 05.03.17 at 14:30, Catch-all, null, Leave a comment Permalink

Well, this is my first nonplatonic blog post.