Blessed by the Potato

Another bad week for exercising, but the diet was at least a bit better.

A NYT story on credit scores suggests that the pendulum has swung too far in the states, and now it’s becoming hard for even borrowers with decent credit to get a loan. Some mortgage brokers are lamenting that too much weight is being put on the FICO score:

In fact, FICO scores are not the best predictor. The amount of equity a person has in his home, his debt-to-income ratio, his job stability and his cash reserves are all better predictors than credit scores, according to Dave Zitting, the chief executive of Primary Residential Mortgage, a leading mortgage lender.

Now from what I’ve read I don’t know if I’d say the amount of debt-to-equity is a better predictor, but it’s certainly up there. This just reinforces my earlier point that the line “Canada doesn’t have a subprime mortgage problem” is glossing over the prevalence of CMHC-supported low/no downpayment loans, which while not quite as risky as a pack of NINJA negative-amortization loans, are still much riskier than the “conservative” banking culture played up in the media.

Aside from seeing another source to use to trot out my ongoing argument, the article isn’t all that good. It belabours the point that lending criteria use hard cut-offs sometimes (like here with FICO scores) where the difference between just over and just under the line are too small to be meaningful. Unfortunately, life is full of such arbitrary cut-offs: for instance, if you have $499 in your account but write a cheque for $500, it’s bounced all the same as if you had nothing in the account. These cut-offs can help protect the larger system (e.g., the bank) from bad risks from decisions made throughout the organization. Though the article didn’t mention it, there is one solution to the issue: make the cut-offs continuous rather than binary. Rather than someone with a credit score of 620 getting a loan and someone with 619 getting nothing, scale it in so that someone with a 650 could get a full loan right up to their debt service level of say 32%, while someone with a 620 could only get 20-some percent, and scale it down to zero over a wider range… But that’s nitpicking the point.

Rogers has tightened up their download limits again, just days after Netflix announced it was coming to Canada. Whispers of market manipulation to shut out a competitor to their own video-on-demand service arise.

An older article from the CBC goes over some of the basics of download limits, and some of the anti-competitive issues. As you all well know, I think the companies are BSing us here — first off, their “average user’s usage” figure hasn’t moved in years, despite the prevalence of things like streaming video in the last few years. I think it’s probably way out of date now, likely a factor of 10 too low.

the company says the caps were necessary because between five to seven per cent of its customers were using more than 80 per cent of its bandwidth, thus slowing service down for everyone.

This is an argument trumped out often in favour of caps or limits, but what does it really mean? Years ago, during the first round of ISPs cracking down on heavy users, these sorts of arguments were used to cut “abusers” off… but this sort of relationship is just a feature of how humans distribute resources. It’s the Pareto principle. Plus of course, data transfer is very cheap, on the order of cents per GB, yet the overage charges are $2/GB. Even with a healthy extra put in there to act as a disincentive, this is clearly a massively profitable area for ISPs, way beyond the costs of data transfer or economic disincentive. The real issue they often complain about with their networks is peak usage, i.e. time-of-use (especially Rogers’ architecture). Yet they’ve taken no steps towards time-of-use billing, even though that would make more sense.

On the StarCraft 2 front I ended up using one of my guest passes since it’s just getting ridiculous that they can’t solve the account problem that’s kept me from playing. I finished the single-player campaign, and found it quite short. There were 26 missions (Terran and a few Protoss side missions), which compares well to the number of missions in the first StarCraft. However, I found the missions to be very fast and small-scale. I don’t think there was any one level that took me more than 30 minutes to clear, whereas I remember at least one level per race taking over an hour in the original as you had to carefully pick your way through the enemy forces and win by attrition sometimes. Detaching the single-player a bit more from the multiplayer did add some neat options with unit upgrades and mercenaries, as well as a greater spectrum of units (e.g., the medic, wraith, and goliath were cut from multiplayer). Still haven’t played multiplayer though. I tried calling the support line a few times last week, only to find that I couldn’t even get into the holding pattern since the queue was full. When I did finally get through, though the fellow was nice, the problem didn’t get solved. I sent an email right away on release day, and found it ironic that the message telling people to call again since the hold queue was full suggested emailing support instead. Finally, 6 days later, a rep has gotten back to me, and after some back-and-forth going through the motions of trying steps that everyone in the support forums said didn’t help (and that I already tried on my own), it looks like I should get my account fixed tomorrow (8 days after release). Update: Just got in, woo-hoo! Now I’m too tired to play though and have to go to bed…