As I discussed in my last post, a tie could happen in the upcoming US Presidential. And things get very complicated, controversial, and interesting in that case. In the one scenario I examined closely, a tie leads to the Presidency going to the Republicans.
ABC News reports that a possible tie is being taken quite seriously by the Democrats, reporting that "according to the Obama campaign, there are at least 19 plausible scenarios in which Obama and McCain could each have only 269 electoral votes after Election Day -- one short of the 270 needed to win the presidency."
I was fiddling around with the LA Times' cool interactive US electoral voting map today. Applying what looked reasonable based on past outcomes, the first scenario I came up with was a 269 to 269 electoral vote tie for the presidential election. And apparently there are many other tie scenarios thought possible by others.
I am starting to get a deeper understanding of what defines political parties and their platforms. A lot of that insight comes from thinking about the way computers automatically generate decision trees to classify data.
Imagine you have a set of questions. They could be about anything that people care about. Do you like pineapple on your pizza? Do you support free access to abortion? The only important thing about the questions is that answering them consistently maximizes the seperation of people into groups within which answers are as homogenous as possible. The set of questions that achieve this end is what we call a "party platform" or a "political platform".
If the number of members in each party platform group are about the same, you have a natural two party system. There are enough supporters to elect either party, and luck or political winds will dictate the actual outcome. If there is a big difference in membership size, evolutionary forces will drive change.
Big parties will tend to break apart if a smaller branch of the decision tree is more homogeneous than the larger party, and also large enough to assume more power going alone than within the existing party system.
Parties that are too small to win power will change their platforms to maximize their inclusion of members from other parties. They will be limited in this effort in their need to retain internal homogeneity sufficient to stand out in the political landscape.
Whenever you run a machine based learning algorithm to create a decision tree, the outcome is not entirely predicatable and depends to some degree on the order in which questions get applied. Is the first branch in the tree "Are you Catholic?" or is it "Are you a minority?" or "Are you middle class?". When many candidate questions achieve the same level of separation into large evenly sized groups, it becomes a bit arbitrary which one you choose. But the choice greatly affects the resulting set of questions used to create the resulting homogeneous group - the resulting political platform, and who gets driven to most associate with what party.
The fun outcome of this thinking is that you could create artifical political platforms using machine based learning. You could create nonesense parties based on trivial questions like "do you like pineapple on your pizza?" Or you could take all the most pressing questions of the day and generate new political platforms whose members are more homogeneous in their beliefs (less fractious) and at the same time large enough in membership to win a political contest.
Who would win? The CART party? Or the Republican party?
Down the street from my house there was a Vietnamese funeral going on. These extravagent affairs are quite common public displays. It was quite different from a western funeral, with loud music and singing transexuals, and a large crowd of gawkers. I think the singers were a bit unusual, even here.
SinglePostcard is our latest Facebook application. It lets you create postcards using your own uploaded or online photos and have them mailed anywhere in the world.
- High resolution postcards are printed with no branding, advertising or logos.
- The regular price will be $2 to anywhere.
- Early adopting users can send two postcards free, anywhere in the world. Currently this is a limited offer for the first few hundred users.
-
A unique feature of SinglePostcard is its social discounting system. The price permanently falls to $1.20 for the first eight friends to send a card. But if eight of your friends have already sent a card using singlepostcard, you get no discount and will always have to pay the full $2. So don't get left behind, sign up and use the app early!
-
Recipients of cards receive a 20% discount code that they can use to send a reply or to send a card to another person. The discount code is printed discretely on the back of the card, but can be turned off if you find this intrusive.
The URL should be at */singlepostcard on Facebook. Until the launch time (see below) we'll be doing some final testing of the site so this URL might not work until then.
We plan to launch worldwide by 8 pm Sunday (San Francisco), 11 pm Sunday (New York), 4 am London (London), 1 pm (Sydney).
I currently plan to go to the first Barcamp in Cambodia at Phnom Penh on Saturday September 20th -- in less than three weeks. It should be a cool breakthrough event for a place where all free or intellectual thought was quite recently cause for immediate execution. A flickr stream of pictures is developing for the event using barcampphnompenh (the official tag). You can expect to see media posted elsewhere, such as youtube, using the same tag.
It is almost universal, the more expensive the hotel, the more likely they are going to charge you a silly amount of money for Internet access. The most frustrating part is that hotels will say they have broadband Internet in their rooms, but rarely say if they charge for it, or how much. Typically, one might pay $200 for a hotel room, only to find there is another $20/day charge for Internet access. And that access is often poor.
For example, I'm staying at the "Swissotel: The Stamford" in Singapore right now. It is a top hotel -- one of the tallest in Asia, with basic rooms in the $250/night range. They say they have broadband in the room. When I arrive in my room, I find I have to pay $19.87 (US) per day (plus exhorbitant tax). And the speed, especially upload, was nothing to write home (blog) about.
Sometimes you do find a really nice hotel, and the Internet is free. Those are sparkling gems. One I found recently in Thailand was the Bel-Aire Princess in Bangkok. At $109/night it had free broadband Internet. It was such a pleasant room and experience, I spent almost an entire day in the room working and surfing the web.
I think the world needs a website that is dedicated to making the hotel Internet situation a lot clearer and more transparent.
In 1991, when I started my first real job, you could tell senior management dinosaurs by the way they communicated. They would hand write memos and give them to their secretaries to enter into Word and distribute on paper. We worked in an organization with thousands of employees, all with computers and a single email system, but they couldn't type and they didn't think something was real unless it went out on paper with their initials on it.
The new staff immediately took to email, and we put it to good use organizing an endless stream of open house parties around town. If you had email, you were ok and welcome.
Today, voicemail is the new hand written memo. If you communicate with people via voicemail, and expect people to leave you a message, you almost surely don't get it. You're almost surely The New Dinosaur.
Arrington has it right: Think Before you Voicemail.
Work started in our office at eight in the morning. But before nine, the power had failed. And it was not back on within a few minutes, as had been the case in the past. So I decided to put into effect my coffee shop plan for business continuity. Later we would find out that the government planned to keep the power off until four in the afternoon; this would be one of Saigon's summer power outages I had read might happen.
I designated the day one for research for some -- that means reading technical books (not using electricity or computers). And with a laptop, and an iMac (they are surprisingly portable), we moved to a cool coffee shop to continue working for the day. We got off to a bit of a slow recovery because the coffee shop had to get its power generator working. But eventually we were sipping coffee and fruit drinks, eating some great food, and listening to Jazz while we worked from our new office away from the office. I think the day was almost as productive as usual, once we got going, and I'm sure we all had more fun.
I chose to set up in a new coffee shop called "Squared", located at 5 Ly Tu Trong in District 1 of Ho Chi Minh City, Vietnam -- just across the street from the popular club, "Lush". The coffee shop is owned by a tobacco company, which did not exactly endear it to me. Although I must admit I did think it was rather interesting that they have a tobacco bar where they will make a pack of cigarettes for customers using a blend of tabacco and flavour of their choice. The reason I
chose this place was because it has good food, pleasant Jazz and Lounge music, and most importantly, good Internet. Today I measured about 350 kbps in both directions with Silicon Valley, but in past days, I measured about a megabit per second of bandwidth.If power failures like this happen frequently, I might buy a back up generator. But this coffee shop solution seems to work well.
on Vietnamese Coffee