With the upcoming IPO here at Shutterfly, I naturally wanted a way to create a bit of a prediction market around it(some would call that gambling, but euphemisms are more fun). I’m most interested in how to set up reasonable market to predict the end of day price.

It’s obvious that a simple “every entry is $5, winner take all, closest to the number” is a terrible terrible idea. If you don’t understand why, think about what happens when I pick a number and then immediately get straddled by two other people picking numbers 1 cent above and below me. You could add minimum spacing increments(which is what we did with $.05 increments), but that only solves a part of the problem.

So what’s a better way? I thought of doing an auction market on every $1 increment in pricing. In summary, it’d be “bid on whether price will be $11-12, $12-13, $13-14, … every winner for each range has to pay up, and the correct price range takes it all” . It’s a bit more work, but seems a lot more fair.

I think that is pretty good. But I’m not sure if that’s efficient enough. Choosing the size of the ranges might be a bit tricky. But you can make ranges flexible by saying that any range can be subdivided by its current winner into two equal, half priced segments. And any current winner of two adjacent bids can consolidate the two ranges into one range as long as its new price is at least the sum of the individual prices. This is all nice, but the whole thing is still pretty complex. Maybe I’ll try this out for my next IPO.

It’d be good if there were a scheme which didn’t require individuals to track the whole market. If someone knows of a simple scheme which was just “I’ll buy in for $X and make a prediction of $Y” where the early predictors don’t get obviously and massively screwed tell me.

I’m an uncle yet again. My brother Tim had two healthy twin boys early this morning. The names are Robert and Jonathan. The family is doing well. Congrats Tim.

Here’s what the new government of Thailand said after their recent military coup.

“The armed forces commander and the national police commander have successfully taken over Bangkok and the surrounding area in order to maintain peace and order. There has been no struggle,” the new leaders said in a statement on national television.

“We ask for the co-operation of the public and ask your pardon for the inconvenience.”

If only it were that easy to do here in America. They didn’t even have to resort to violating human rights with torture or imprisonment without trials.

Wow this article really scared me. I don’t believe in a national election fraud, but the evidence makes me think that there has to be something wrong in Ohio. I’m impressed with the level of dirty politics that Republicans are going to. Lying and questionable ethical behavior are just not good enough anymore. Now they have to go to outright felonies.

In another political rant… I hate how people think of terrorism as a real threat to America. I’m not saying that we should do nothing, but spending money on security always has a low ROI. It’s about risk tolerance. We should just accept that these things can happen. It would suck yes, but we’ve spent trillions to fight the war on terror. Couldn’t we have saved a million lives with that money instead of maybe preventing a terrorist attack?

Ugh. I prefer not to think about politics anymore, it just gets me depressed. I want the Democrats win some seats in the November election, but I am going to stay cynical until the election is called.

Lucas has had a fever since yesterday and that kept us up most of the night. And I made an unfortunate mistake at work which caused me to have to work pretty late too. I only got about 5 hours of sleep per night this week. I hope I can catch up this weekend.

This morning on the drive in to work, I saw a guy in the next lane wearing a yellow rubber Livestrong bracelet light up a cigarette.

Lucas is doing well. He can crawl everywhere now. He’s just starting to stand. I think he can stand for about 3-5 seconds when he’s really trying. He’s still a very cute baby, but he’s not always sleeping through the night. Tracy put together a couple of Shutterfly’s new 8×8 and 12×12 photobooks that I got with the employee discount and they look very good. The paper and print quality is excellent.

I had a couple of thoughts on interesting things that could be done with some smart software.

1: Process migration instead of data transport. In today’s distributed computing environments, given a pool of storage servers and a pool of application servers, one of the challenges is figuring out the best way to get the data to the application server. I think someone could make a cool product if they could turn that around by instead of getting the data to the application, move the application to the data. If the data is large and size of application is small, why not just move the process instead? Several languages support it and there are probably extensions to Java that can do it too. It’s not too well suited if the application needs access to many disparate data sources concurrently, but for a number of applications, it might make sense.

2: Click traffic arbitrage system. I’m sure someone has or is doing this already. Buy a click on Yahoo for for $.10, sell it on Google for $.12. Not very interesting from a computer science perspective, but easy money.

3: Non-uniform disk access raid. I have a PC, I’d like a raid array which I can stripe and/or mirror to which provides more intelligent performance than just running as fast as the slowest member of the array. That way as, newer, faster, bigger disks become available, I can just add one drive to the array and get better performance and capacity without having to upgrade all the drives. There’s some complexity because you have to understand the costs(speed,size,failover risk) of working with each separate disk and manage them accordingly.

Anyone wanna take on one of these?