Archive for May 2004

Off to New York!

Back in a week.

That’s not what I meant, really!

I work with the technical teams at both NetZero and AOL regularly[1], and it’s surprising how accurately the technical competence of each company’s staff reflect their own user base. NetZero’s team is hard core, whereas AOL’s is usually not on the ball.

Anyway, a guy at NetZero said something today that I didn’t think was accurate, but I consulted my good friend Google to make sure. I didn’t notice the double entendres until the last iteration in my choice of search terms:

  • dirty anon pages
  • dirty anonymous pages
  • dirty private anonymous pages

So all you site owners, don’t always assume the worst when you see search terms like these in your referrer logs. 😛

[1] Both companies use our technology for their accelerated dial-up solutions. NetZero here, AOL here.

Something for which I’ve never heard a satisfactory explanation…

I’m no prude, but damn. That is one spicy meatball.

Ever see that movie, Summer of Sam? John Leguizamo’s character is married to Mira Sorvino’s character, and he has this hangup where he’ll only shag her in the bedroom, with the lights off, and in the missionary position. Meanwhile, he’s off doing the nasty behind her back with any willing participant in all kinds of hardcore-porn-like ways. His explanation is something along the lines of wanting to keep her pure and unsullied. That’s an extreme example, but many (most? all?) males seem to have at least a little bit of this tendency. Take that Washingtonienne link up above. Bunch of married senators/politicians doing similar things.

A less extreme example a little closer to home. Bonnie sometimes plays this game where before we go out, she asks me to pick what she should wear. More often than not, I miss the mark. My natural inclination is invariably to pick something too conservative. I have to consciously try to pick something in the opposite direction in order to counter this tendency (which usually results in something equally inappropriate, but on the other side).

It’s not like I don’t appreciate when someone is dressed appropriately and looks good doing so. Like any warm-blooded male, my gaze lingers on girls that are easy on the eyes. So why, despite all of this, is it not my natural inclination to dress my girlfriend in this way when given the choice? I suppose you could chalk it up to an insecurity – something like, if she looks too nice, someone might steal her away from me, so I’d better make sure she looks frumpy whenever possible. Perhaps, but if that’s true then it must be something that’s pretty deeply engrained, like a bad reflex. I consciously know that it’s incorrect, yet it still happens.

Anyway, I’m not sure I had much of a point. I’ll just leave you with something I came across while reading other blogs about that Washingtonienne girl. Never heard this term before, but then again I’ve never been down with the cool kids. Funny nonetheless. Butterface.

Intel and BMW

Shows how much I know about cars, but I saw a BMW 8-series coupe on the road the other day in North York, and I thought it was new. Nope. I didn’t even know BMW had even-numbered lines, apparently there will be a 6-series soon.

Speaking of BMW-esque numbering schemes, most of you will probably have heard that Intel is going to be transitioning to just such a scheme for their CPUs, using the same three digit model number as BMW, separating their offerings into 3-series, 5-series, and 7-series lines.

Odd though, unlike cars, CPUs don’t come out with a refreshed lineup every calendar year, so it isn’t possible for Intel to reuse model numbers. Am I the only one that sees a problem once Intel releases 100 CPUs for each line? Maybe they’ll switch to a 4-series, 6-series, and 8-series. Or maybe they’ll finally drop the Pentium name and have a Sexium/Hexium 3, 5, and 7-series.

In any case, Intel’s new numbering scheme definitely loses information, and I’m not sure that it transfers well from the automobile world. Sure, you know the higher numbered CPUs are better, but they have so many offerings for each 3/5/7 line that it just becomes confusing (at last count it was in the 10-15 range when including both desktop and notebook CPUs).

When a car has a higher number, you can say with virtual certainty that that the engine is larger (more cylinders, larger cylinders, etc.). Engine size is the primary factor in determining performance, the same way MHz used to be the primary factor in determining performance for CPUs.

What are you going to say for a higher numbered CPU? The higher numbered ones won’t necessary have a higher MHz, the whole point of this numbering scheme is to downplay the importance of MHz. Instead, performance comes from a combination of closely-weighted factors, of which MHz is just one of them. I can’t picture the common folk talking about or understanding things like cache size, cache associativity, FSB speed, pipeline depth, branch prediction table sizes, number of functional units, etc.

I think the scheme’s going to blow up in their face if they don’t change the technology back so that it has a single overriding factor that determines performance. You have to let the common joe brag about having a “bigger engine”, then let the technology buffs fight amongst themselves over the less significant factors.

They’re probably betting on the multi-core thing (i.e. more than one CPU in a single physical package). They’d better make sure that performance scales close to linearly as you add more cores, or they’ll be back in the same MHz boat they’re in now.

Will Google for Food

Size Does Not Matter

After skimming through Google’s prospectus, I get the same uneasy feeling that I get when someone asks me for handouts. Allow me to explain. It’s easy to lose the forest for the trees when talking about things on a large scale, so let’s examine a microcosm instead.

Say you have this wicked idea for a new widget that you’re sure everyone and their dog will want, except that you’re flat broke, so you can’t even build a prototype. You convince your parents, relatives, friends, and anyone else that will listen to you to chip in for your starting capital. Eventually you raise enough and get to work. Now some questions.

All those people that gave you seed funding… do they have a right to periodically check up on you to make sure that you’re not just spending their money on lavish toys for yourself and living like a king instead of actually building the widget? In turn, do you not have a fidiciary responsibility to ensure that you are maximizing their investment, finding the best way to build and sell enough widgets that you can give a return on investment that matches the risk they took for putting their neck out for you? Of course you do.

The stock market is no different. Its purpose is to match up those that have money (investors) with those that need money (companies). The companies on the market are simply at a later stage in life (versus the example given above), so they are usually quite large in terms of both employee count and annual revenues. Despite their size, the management of a publicly traded company has the same responsibility to its investors. If it becomes clear that management is not demonstrating this level of responsibility, shareholders have a right to invoke any necessary changes.

Don’t like where you work? Blame management. Blaming the stock market and shareholders borders on the comically absurd.

Back on Track

Back to Google, the reason it feels like someone asking for handouts is because of the structure they have chosen. They’ve split their common share pool into two classes, Class A is allowed one vote per share, Class B is allowed 10 votes per share. The two classes are identical in all other respects. The Google founders hold all the Class B, the Class A is being sold to the general public. They intend to raise about $2.7B in this initial offering.

The proposed structure essentially says to the purchaser of the Class A common, “Please give us $2.7B, an order of magnitude more cash than we currently have on hand, but you can’t say anything if we decide to fart it all away on gourmet meals, massages, on-site physicians, and Gwyneth. Because you know, our employees need this kind of thing. Trust us, we went to Stanford.

In other words, they want your money, but all you get in return is their word that they’ll be responsible with it. Sorry, not convincing enough, there needs to be something more to balance this additional risk that they intend to thrust onto the holders of Class A common. Something like a dividend, perhaps? Not a chance. For all intents and purposes, the Class A common are preferred shares, only without the dividend, and without the seniority in the case of a liquidation event.

The prospectus cites companies such as Berkshire Hathaway and The New York Times that have similar two-tier common share structures. What they fail to mention is that the former is essentially a holding company where you pay the Oracle of Omaha (Warren Buffet) to manage your money, and the latter (supposedly) requires such a structure to maintain the quality and impartiality of its news reporting. Google doesn’t have a reason even remotely as good as these two.

So it’s no wonder that they’ve decided to sell their offering as an online Dutch auction, under the guise of “bringing their long-awaited IPO to the common-folk”. The truth is, with the terms they propose, they’d have a hard time selling to institutional investors, so they have to milk retail investors instead.

Imagine you managed the portfolio of the Ontario Teachers Pension Plan[1]. Would you like to be the one to tell your members, “Apologies, some of you won’t be getting a pension this year. We bought $500M worth of Google Class A, and their management arraged it so that we had no control. Then they went off and made some really really bad business decisions. How were we supposed to know this would happen, Larry Page and Sergey Brin gave us their word. Their motto is “Don’t be evil” for christsakes!”

I’ll stick to companies that aren’t afraid to remain accountable to the people bankrolling them, thanks.

[1] The OTPP has some $87B in assets. They didn’t get that big by taking on risks that don’t have a correspondingly large enough reward.