Just received my replacement board from ABIT, so I'll be swapping it in and hoping for the best. For those interested, the RMA process is totally electronic, and zero hassle, so despite my annoyance at the lack of general support, it looks like it may work out. I guess its cheaper for them to handle RMAs versus staffing up a larger support team.

One of the things that I try to avoid when debugging a problem, especially one to which I don't have real insight into, is assuming that because two things co-occur, one must be the cause for the other. More often than not, the two things are actually both caused by a third, unseen event. However, when my Roadrunner email (which I rarely use) suddenly started telling me "account disabled" immediately after I had a new digital phone line installed, it seemed to be too coincidental.

The real fun was trying to figure out how to get Time Warner to listen to me without having to sit in a long tech support queue and argue with some level 1 tech about why it wasn't working as they walk me through unplugging and replugging my cable modem, forcing me to narrate my pretend actions so that they will finally transfer me to the next level.
I have learned that you will always get more leeway if you buy something, or ask about available services and their prices when you actually need tech support. This works for two reasons, in my experience:

1. They have a more customer-service oriented mindset, and so they want to make you happy.
2. They tend to be less technical, or at least have fewer notions about being a technical person and so they will listen when you offer up a problem diagnosis.

I remembered that I had asked for voicemail on the line but they did not add it for whatever reason, and this provides a clear "in" to the sales line. I try the tech line first just to see if I can quickly resolve it, but the wait is > 45 minutes. I then try the sales line; I get a person right away and I explain that I want to add VM. No problem, they can add that for me.

Me: Oh, and BTW, my email account stopped working right after the phone was installed.

Sales Guy: Oh really, let me look at that...yeah it looks like it got confused here. Let me fix that for you.

Me: (astonished silence)...uh, thanks

10 minutes later my VM was active, and my email was back online. So anyway, the moral of the story is: to get good tech support, tell them that you want to buy something.

I have two WRT54g routers (well, one 54g v4.0 and one 54gl since they messed with the 54g line after that version), both "upgraded" with the OpenWRT firmware. One of the things that I use them for is to bridge them together over WDS so that I can have a bunch of computers talking to each other over a wired link upstairs in my office, and then have that router bounce traffic destined for outside or other computers downstairs over the WDS link. I had it seemingly set up okay, but I started to notice some weird effects, which I've since learned had to do with my failed understanding of what was actually happening.

• All wireless clients would only connect to the downstairs AP. I figured they would talk to whomever is closer, and this defeated the secondary purpose of this set up, which was to provide better coverage for wireless clients. Since the WRT54g has a much better set of antennas, it seemed like it would be a win for clients upstairs, not to mention that they would "think" they were connected more often, which is just as good.
• Occasionally I would have weird wireless dropouts. I figured that this would no longer happen since there were multiple access points. Instead I expected to see the wireless signal staying strong, but the packets being dropped. That's how I discovered the above.
• I would get duplicate packets when I pinged things from a wireless client. I attributed this to just an odd side-effect of my setup, and it didn't seem to affect anything major.

Well, it turns out that what I was doing was a little bit screwy, and I'm still not even quite sure why it worked. First of all, I set it up in client-bridged mode, so the upstairs router was not acting as an Access Point. In my mental model of the situation, I assumed that being an access point and being a bridged client were incompatible and that the latter subsumed the former. Even weirder, I was using the wrong mac addresses for the WDS connection, so I think I was actually entering the network at the wrong level and I think that was what was causing the duplicate packets, although I'm surprised WDS worked at all.

My real problem is that I failed to understand that using WDS and being a client are orthogonal properties. You can turn on WDS in all your routers, and still let them all be access points, or let them be clients. As access points, they all accept wireless clients, and route traffic to each other accordingly, which is what I wanted. When I switched the upstairs box to be an access point, it suddenly stopped routing packets to the downstairs box, and that's where I got the idea that they were linked together. Somehow, when I set the box up as a client/repeater, it worked, even with bad MAC addresses. Once I fixed the mac addresses to correctly use the wireless devices, it worked as expected, in access point mode.

Now if I scan for access points, I see the one with the stronger signal, and my wired boxes upstairs can all get out through the WDS connection. For anyone who has played around with this, make sure you use the MAC addresses that you see when you say 'iwconfig eth1' on your routers, and not the one that shows up on the status page. That was my main mistake. This was also a great example of how you can fit observed symptoms to an assumption of functionality that proves to be totally wrong.

I hate going to the DMV for the same reason I hate going anywhere where lots of different transactions can be performed, and there are lots of prerequisites for any of them. Here are the core problems:

1. It seems as though the point of the DMV is to make sure you leave having failed to accomplish what you came there for.
2. Since they do this all day long and you don't, they treat you like an idiot if you can't remember the 37 things you needed to fill out and bring to register your car.
3. The employees seem to take a perverse sort of pleasure in 1 and 2.

I have to give them some credit: they seem to have added a new initial person who acts as a early rejector by telling you all the things you failed to bring or fill out so at least you don't sit there for 20+ minutes before getting a DMV teller to cackle wildly when you present your out-of-state license transfer form without a notarized signature from your godmother's cat.

There are other problems though: hours are haphazard, at each location some transactions can be performed and others can't, and the forms are dense with DMV jargon. I think there are some simple things that could be done to fix the DMV and make it so that it's no worse than going to the bank or some other semi-favorable transaction.

1. Post huge signs in the entrance with the 3 or 4 most likely transactions (registration, licensing, titles, road test, let's say), with a checklist of items, and a step-by-step procedure. This goes for any place that has a set of implicit scripts to follow that they seem to keep hidden for whatever reason (I'm looking at you, Post Office).
2. Offer these same things on the web. If you got to the WI DOT site, you get a zillion pages with somewhat conflicting information. Really it has to do with the fact that there are really different categories that you might belong to that are actually well-defined, but it's hard to know which one you fall in to. In addition to the giant checklists/process steps, offer a simple way to know which category you fall into. Having moved here from MA with valid plates, am I doing a plate transfer or not? I had to ask the guy at the counter to find out (No). Turns out I am essentially in the same category with people who had just purchased a new car.
3. Streamline the delivery by offering either every service at every location, or only one service at each location. Right now, you have to look at each place and then try to guess if the thing you want to do is offered at a particular location or not. Also, please don't just be closed on one random day each week. That means I have to check the website every time I need to go just to make sure it isn't that day.
4. I know you want me to do everything by mail, but people have questions, lots of questions. You need to offer some kind of web-based or phone support if you want everything to be done by mail. It will save time and money overall.
5. This one is the big one: an attitude change. People come in there with little or no understanding of what they need to do, and generally a relatively simple request. Why must we be made to feel like criminals and idiots? There has to be a way that someone could come in and start collecting user feedback about their experience because I know I'm not the only who feels this way. Their job is to help us get things done, not to prevent us from getting things done.

I think the key lesson is that municipalities could do a better job delivering services if they changed their default policy from DENY. I know that they are trying to cut down on fraud and all, but the vast majority of citizens are trying to do something perfectly legitimate. There has to be a better way of preventing illegal transactions than making the process increasingly daunting.

I attended the first BarCamp Milwaukee yesterday, and I think I had the same reaction as many people walking in there: "I can't believe this tech community exists in Milwaukee", which is just silly. I know some people came from other cities and even states, but I think that it speaks to the weird mismatch between perception and reality in terms of the tech community here. I'm hoping that one thing that can come out of this is an ongoing set of meet-ups, possibly continuing at Bucketworks, rather than having this one-time (although fabulous) event, to keep and build momentum.

I attended 2 sessions and led a third. The first session, about Ruby on Rails was great for me (I was the guy asking way too many questions) since it was something that I'd heard a ton of great things about but hadn't had time to really investigate. The second session was about gadgets, and I really liked the discussion regarding the open-source vs. proprietary software model in the work of small devices, as well as looking at why certain devices failed or succeeded. I led a session about Linux in everyday devices, focusing on how things become hackable with Linux inside, and whether that is a good or bad thing for consumers and for businesses. It was actually an interesting follow on to the discussion that started in the Gadgets session.

Anyway, kudos to the organizer and I hope this is the start of something bigger here in Milwaukee.

I moved recently, and my Linux server started having trouble when I set it up at my new place. Specifically, yum and firefox kept crashing out oddly where I never had any previous errors. I assumed it was a memory problem, so I ran memtest86 on it for a while, and I got a whole slew of errors on Test 7 where it reads and writes random values. I also tried memtest86+, which would just hang when it got to a certain set of operations. I tried removing each memory stick separately and rerunning, but the errors persisted. At that point, I had to assume that something had gone wrong either on the CPU or the motherboard somewhere.

Long story short, the machine was 2 years old so I figured I was due for an upgrade anyway and purchased the components for a new Pentium D-based system, which I figured was the best performing thing in my price range. I could assemble an ABIT AW8D mobo-based system including 1GB RAM for about $400, which is nice. I threw in a new video card since the mobo was PCI Express. It all arrived last week, but I got this funny feeling when I assembled it that it wasn't going to work, and my "this-ain't-right" detector is usual pretty accurate. I don't know what it was, but sure enough, I apply power and the thing gives me a series of beeps (1 long, 3 short) which indicate a keyboard error, apparently. 2 PS/2 keyboards and a USB later, still no change.

Oddly, the POST code readout shows 8.7. which could be a CPU voltage problem, so I don't know who to believe. I thought maybe my old 350W power supply couldn't handle the load, despite the manual's claim that 300 was sufficient, so I swapped in a 500W Seasonic S12 which is rated one of the best by Tom's hardware. Still no change. I've appealed to the distance debuggers out there on the Abit forums but haven't gotten any responses. Might just be RMA for me...

If you have any suggestions, please leave them in the comments.