Fatal Attraction

For a long time, I’ve told people that I avoid online multiplayer games (except chess) not because they’re uninteresting but precisely because they are interesting and I didn’t want to get addicted the way so many others have. Well, I guess I’ll have to find another addiction-avoidance strategy, because I’ve started playing Dofus recently. Apparently it’s pronounced doe-fus, not dew-fus, but I’m sure the latter pronunciation is common among players of more “serious” MMORPGs (as though running around killing imaginary dragons with imaginary swords isn’t pretty silly in its own right). In fact it’s the un-seriousness of Dofus that I find attractive. Most of the names are palindromes, anagrams, puns, or funny references, though many are French and so less obvious. There’s good entertainment to be had just by reading the descriptions of items being advertised on the trading channel of the in-game chat system. The cartoony drawing style might not appeal to some, but it serves well enough, and I actually like the turn-based tile-based combat system. It’s less frantic than most games, more like a board game than a brawl. These factors are also intimately tied to the fact that the whole thing is Flash-based, which makes it playable on just about any platform; since I use Linux day to day and don’t care about pretty pictures all that much, it’s a tradeoff I thoroughly applaud. (Wakfu, from the same company, seems to be essentially the same game with better graphics at the expense of becoming platform-dependent.)

One thing that’s a bit sub-optimal about Dofus is a relative dearth of information for noobs – by which I mean particularly people who are not only low-level but entirely new to the game and perhaps the genre. Even the most basic guides seem to assume that you recognize the significance of certain game-play elements, though there’s no real reason to suppose a true noob would know that. As one noob to another, then, there are some tips.

Creative Digital Camera Uses

Here’s a link to ten less-than obvious ways to use your digital camera (via J-Walk Blog. I really like the parking-spot and rental-car ones. My own favorite non-traditional use of a digital camera is to take pictures of the white board at the end of a long meeting. I don’t do it often, but it sure beats copying the content down by hand, on the occasions when the end result is worth saving. Besides, it always bugs me to see “Do Not Erase” on something that was designed to be erased, when the room’s purpose might be to contain a large erasable writing surface. I always want to draw a little “no littering” sign next to such notices.

This Blog = 203

This is hilarious: HTTP Conversation Codes. I want 406 on a sweatshirt, and 410 on a resignation letter.

And yes, it is possible to assign “racier” meanings to most of these. This being a family blog and all, I’ll leave that as an exercise for the reader.

SGI Molecule: Imitation is…

At Supercomputing ’08, SGI has announced “Molecule” – a research project that could put 10,000 cores in a single rack. According to Shahin Kahn:

Analysis of applications and market trends shows that several customer environments can potentially benefit from the combination of significantly higher density, memory bandwidth, memory bandwidth per instruction, and performance per watt

Well, duh. Welcome to the party, guys. Some people did that analysis years ago, and have been shipping products based on exactly those ideas. No “research project” about that, no “could” do it a couple of years from now; just products that customers can buy today. It’s not 10K cores in a rack – it’s 5832 in approximately a double rack – but that actually includes things that SGI didn’t include in their calculations. Good luck putting your 10K processors and memory for them in a single rack. Oh, and what good is a drawer full of energy-efficient processors with massive memory bandwidth if they can’t talk to one another? Now you’ll have to eat some of that space and destroy some of that power efficiency with a nice fat Infiniband switch or ten, and you’ll still have lousy communications compared to your processing power and memory size/bandwidth. Lots of processors sitting around waiting for messages means longer job completion times, and time has to be factored into energy-efficiency metrics to be meaningful.

Hitching this idea to the “Atom” brand is good marketing, but is it good engineering?

Fun with Bugzilla

Some engineers will try just about everything to make a bug (and the user who reported it) go away, except for actually analyzing and fixing the root cause of the bug. Unfortunately, some projects seem to count any contribution to a bug report as a sign of productivity even when it distracts from finding a real resolution and alienates the user. Yes, I have a particular project – and even a particular support person on that project – in mind here, but those who don’t already know who it is will probably recognize the pattern anyway.

  • User My system just exploded. Three people had to be taken to the hospital. The last thing before the explosion was a message in the log from your code that said, “Something’s wrong – about to blow up the system.”
  • Support Guy We need more information to debug this. Please run our diagnostic program which will collect relevant information such as config files, logs, credit card numbers, etc. and then attach the result to this bug report.
  • User I tried to do that, but it said the attachment was too big.
  • Support Guy Yes, it has to be under 5MB even though the program routinely produces output larger than that. And our web server will mangle it if it’s compressed.
  • User I manually edited out a lot of repeated warnings – wow, you guys sure put a lot of spam in my logs – and attached what’s left.
  • Support Guy I see from these logs that one of your servers has a name starting with J. As you know, that’s an invalid configuration. Closing the bug.
  • User I changed the server names and the problem happened again. Fortunately, we had evacuated the room this time before we ran the test. New logs are attached.
  • Support Guy This looks a lot like bug 12345. Closing as a duplicate.
  • User I applied the patched from bug 12345, and it happened again. We lost an intern this time, but that’s OK.
  • Junior Developer Here, try this patch.
  • User It didn’t explode this time, but it did melt down. We have a big hole in our floor, and only one wall left standing.
  • Junior Developer Well, that’s progress. This seems to be another instance of hitting a deadlock, because we ran out of threads to handle requests, and then overreacting when one of those requests times out. This patch should fix the case that you hit (while leaving a thousand others in there and doing nothing to address the fundamentally broken programming approaches that led to them — editor). Senior Developer, could you please review it?
  • Senior Developers #1, #2, and #3 Change the indentation here, use this macro instead of that one, I don’t like the comment.
  • Junior Developer I made all those changes and pretended to run regression tests. Checked in. Say, has anyone heard from the user? His last response was a month ago.
  • Support Guy Nope. Closing as user error. Great job, everyone.

Halloween Pictures – Better Late Than Never

In keeping with my habit of having at least one post in every three be irrelevant to one of my three reader bases – technical, political, and family/friends – here are some pictures from Halloween.

Amy as “Mooch the Messy” (from one of her books) at the kids’ Halloween parade. Most people guessed she was a mouse, a few guessed rabbit, but not one guessed correctly that she was a rat. Even Amy gave up on correcting them.
A cute dog costume at the parade.
Amy and best friend Keira showing off their vampire teeth.
The town party at the rec center was more geared toward bigger kids, but they gave out candy just for trying, so…
Trick-or-treating in town a few days later.

Atmos, Fishworks, Violin, Jaguar, and Green Computing

It never rains but it pours. In the last week, there have been a whole lot of things happening in the tech world that I plan to write about, but if I write about them all at once then the posts compete with one another and if I spread them out then the posts lose currency. Before I pick one subject and dive into it, here’s a quick rundown of recent developments.

  • Steve Todd has been writing about Atmos, previously known as Maui, which is being billed as “Cloud Optimized Storage” by EMC.
  • Bryan Cantrill has been blogging about Fishworks, which is Sun’s latest attempt to get out of the storage-industry cellar.
  • Violin announced a 4TB flash memory appliance which – contra Wes Felter – is distinguished from Fusion primarily by access method and not size or color. Fusion’s differentiator is that they’re right on the PCI(-E) bus, not on the far side of some external interconnect. That’s actually pretty important, even though it would be far more useful if they hadn’t screwed things up by making their implementation specific to one CPU architecture.
  • ORNL announced the new version of Jaguar, the second machine to break the petaflop barrier and probably the world’s fastest computer (SC’08 is next week so we’ll see what the Top500 folks say).
  • SiCortex announced the Green Computing Performance Index to address shortcomings in the Green500, based on Matt Reilly’s proposal.

I plan to write separate posts about Atmos and Fishworks soon, but I’ll just leave them be for now. Hey Steve, since you’re making a series out of this, how about a post on how Atmos deals with global-scale latency and connectivity issues, so I can offer my own counterpoint on that near-and-dear subject? I’m going to focus on Jaguar and GCPI below the fold.

More Fair Than Equal

The Washington Post ombudsperson did a study (registration required) of reporting on Obama and McCain in the last several months. I was actually a bit surprised by how little bias was revealed. Here are some of the most relevant numbers.

The number of Obama stories since Nov. 11 was 946, compared with McCain’s 786. Both had hard-fought primary campaigns, but Obama’s battle with Hillary Rodham Clinton was longer, and the numbers reflect that.

McCain clinched the GOP nomination on March 4, three months before Obama won his. From June 4 to Election Day, the tally was Obama, 626 stories, and McCain, 584. Obama was on the front page 176 times, McCain, 144 times; 41 stories featured both.

Only 626 to 584? Or 176 to 144? That’s nothing; Obama is simply more newsworthy than McCain. For one thing, it’s not news that McCain is running for president; after all, he’s done it before. Just the fact that Obama’s black counts for a bit of newsworthiness, though hopefully that will never be true again. More important than either of these effects, though, is the fact that both campaigns were more about Obama than McCain. McCain talked about Obama far more than Obama talked about McCain, mostly to attack but that doesn’t matter from the perspective of making news. It shouldn’t be a surprise that what the media talk about is what the candidates talk about. The McCain camp made the decision to run an anti-Obama campaign instead of a pro-McCain one, and that alone should result in more than the 7-22% difference actually seen in news coverage.

Moving from news to opinion – separate functions of a newspaper, as all should know – here’s another bit of information.

The op-ed page ran far more laudatory opinion pieces on Obama, 32, than on Sen. John McCain, 13. There were far more negative pieces about McCain, 58, than there were about Obama, 32, and Obama got the editorial board’s endorsement. The Post has several conservative columnists, but not all were gung-ho about McCain.

That’s a more drastic difference but, again, this is op-ed and therefore not supposed to be objective like news reports are. Personally, I think the results reflect the actual merit of the candidates rather well. More importantly, they track the electoral-college results. Should the op-ed page of a major newspaper not be reflective of the people’s own opinions about issues or candidates? Even editorial writers should be honest, yes, and diligent (I’ve criticized Boston Globe columnists enough times for lacking either quality), but fairness does not consist of assigning equal weight despite unequal merit. To make WaPo’s coverage of the election more equal – or “balanced” as Faux News would have it – would require making it less fair in the sense that matters.

Lastly, here’s the most significant type of imbalance evident in media coverage of the election.

The count was lopsided, with 1,295 horse-race stories and 594 issues stories. The Post was deficient in stories that reported more than the two candidates trading jabs; readers needed articles, going back to the primaries, comparing their positions with outside experts’ views. There were no broad stories on energy or science policy, and there were few on religion issues.

Bill Hamilton, assistant managing editor for politics, said, “There are a lot of things I wish we’d been able to do in covering this campaign, but we had to make choices about what we felt we were uniquely able to provide our audiences both in Washington and on the Web. I don’t at all discount the importance of issues, but we had a larger purpose, to convey and explain a campaign that our own David Broder described as the most exciting he has ever covered, a narrative that unfolded until the very end. I think our staff rose to the occasion.”

I completely disagree with Hamilton’s attitude, and even find it objectionable from someone in his position. With respect to the balance between issues vs. horse-race reporting, I think the reporters at WaPo and elsewhere rather sank to the occasion, functioning as entertainers and not the journalists they’re supposed to be – and I think that’s a point on which I might find some support even among those who disagree with me ideologically.

Tips of Icebergs

Today was a very frustrating day at work. I won’t go into a lot of it, but one theme had to do with the assumption that the things we can easily measure must also be the most important things. The problem is that sometimes the assumption is false. It’s quite common, actually, for people to make some improvement in terms of things they can measure – instructions, cycles, messages – and then find that the result is actually worse. One common example is tuning a piece of code for performance, only to find that – while that piece of code in and of itself does perform better – the way it interacts with the rest of the system results in worse overall performance. Even when overall performance is improved, it might be at the cost of introducing subtle bugs in less common cases, decreased generality or maintainability, or other hard-to-measure but nonetheless very real drawbacks.

That last point brings me to the more general problem with this false equivalence between ease of measurement and importance, because it’s not limited to programming. This same mistake is repeated throughout other areas of our lives as well. For example, a great deal of economic theory is in my opinion worthless because it’s based on an assumption that everything important can be accounted for in terms of easily measured dollars and cents. All other “externalities” are routinely ignored, even though effects of these transactions on health and environment and our “social fabric” and even (as we have so recently seen) the viability of the markets themselves can sometimes be very important indeed. Over here are sports fans who obsess over statistics which can never really predict scores. Over there are academics for whom “publish or perish” is but one form of emphasizing visibility over value. In example after example, people fetishize the measurements and ignore the reality those measurements are supposed to represent.

I’m not saying that there’s anything wrong with empiricism and measurement. I’m more committed to them than almost anyone else I know, but sometimes you have to step back and ask whether your measurements are serving you or you serving them. Always ask yourself whether there’s something else you should be measuring instead, or whether you should be measuring something a different way, or whether measurement is really the right thing to be doing at all. Measuring the size of the iceberg above the waterline is always the easy thing to do, but it’s rarely the most important.

Walking to Work

On this drizzly and dreary day, I just have one picture to post. This is from October 27, when I happened to have my camera with me and had to park in the far lot at work. As I crossed the bridge that separates the two parts of “the pond” I noticed that the water was unusually still, so I stopped to snap this.