The Now Economy

Back on March 10th, Bernardo Huberman’s group’s work at HP showed up in the Economist’s Technology Quarterly as an example of internal markets at work. But what commodity should one make a market in?

As the article excerpt points out, electricity is sold by the kilowatt-hour, but computing consists of (at least) three distinct qualities: storage, bandwidth, and processing. Further complicating matters is that privacy and security constraints affect how we allocate each type of computing resource differently. So I don’t know how successful “computons” might be, but we need some handle on that concept to successfully decentralized ‘grid’ computing…

Economist.com: Who wants to buy a computon?

Building just such a market mechanism is the nut that Bernardo Huberman, a researcher at Hewlett-Packard (HP), and his team have been trying to crack. The key, Dr Huberman realised, was to have a system that can allow users to assign different priorities to tasks, to reflect their importance. This rules out any system that would simply give each user a priority without differentiating that user’s many tasks. It also rules out a reservation-style system of the sort that airlines use, since a lot of processor cycles (like aeroplane seats) would end up unused, and the system would not be able to accommodate new tasks as they arose, even if they were extremely urgent. In a grid, it must be assumed, demand is changing constantly and unpredictably, and so is supply (since individual host computers on the grid come and go). Mr Huberman’s answer is Tycoon, a piece of software for computing grids that turns them into a sort of %u201Cstockmarket or clearing house%u201D, he says. Users start by opening a bank account and getting credits. They then open a screen that shows all the available processors, their current workloads, and a price list. Users place bids for various processors, using a sliding price dial that looks like a volume control. Allocation is proportional, so that if one user bids $2 and the other $1, the first gets two-thirds of the resource and the second one-third. If the deadline of one task suddenly moves forward, the user can up his bid and immediately get more processor cycles for that task. As users consume cycles, the software deducts credits from their account. The HP team has so far tried out Tycoon on a cluster of 22 Linux servers distributed between HP’s headquarters in Palo Alto, California, and its offices in Bristol, England. Tycoon did well in these tests, and several amusing animated films were rendered using its system. HP has now given Tycoon to CERN, the world’s largest particle physics laboratory and a hotbed of grid-computing research, for more testing. This is only the beginning, of course. Mr Huberman reckons that Tycoon, in its current form, could run clusters of 500 host computers with perhaps 24 simultaneous users. But the ultimate vision of grid computing is for one gigantic network spanning the globe and accommodating unlimited numbers of users. So a lot still needs to happen. For a start, the metaphor for computing grids as %u201Cutilities%u201D, similar to water or electricity supplies, is misleading, since there is no equivalent of litres or kilowatt-hours. Processor cycles are just one component of computing resources, alongside memory, disk storage and bandwidth. Mr Huberman would like to combine all of these factors into one handy unit, which he wants to call a %u201Ccomputon%u201D (a cross between %u201Ccomputation%u201D and %u201Cphoton%u201D, the name for a packet of electromagnetic energy). Tycoon’s descendants would then help to allocate computons across the grid’s global market. Of course, Mr Huberman adds, that will happen in a different decade.

HP Labs : Research: Tycoon

Tycoon allocates computer resources in distributed clusters like PlanetLab, the Grid, or a Utility Data Center (UDC) using a market-based mechanism where user pay for their usage using a currency. Tycoon allocates the cluster more efficiently than time-sharing schemes, and allows users to change their allocation in seconds.

James Annan, the climate scientist I blogged about earlier, is at it again. He is in the mainstream on the global warming question. AFAICT, that’s pretty unusual for someone who is

loudly challenging others to stake their reputations on a specific bet. Usually, people out of the mainstream use this tactic in order to get some visibility for their unusual position.

The debate seems to have three factions: Skeptics, Alarmists, and the mainstream. The mainstream is represented by the IPCC report (which Annan agrees with). The Skeptics believe that the IPCC report is an extremist statement because it presents the alarmist view and the mainstream together without preferring the less extreme scenarios. There’s some disagreement about the roles, though. Annan thinks that the two interesting positions are those who agree with the IPCC, and those who think it is overly alarmist, while Chip Knappenberger (characterized by Annan as a Skeptic) thinks that he and Annan are in agreement as to the likelihood of the rate of (near) future temperature change, and proposes that they should work together to bet against “the alarmists - those folks who entertain the idea that the IPCC extreme temperature change scenarios are the most probable.”

Annan started this particular brouhaha when he noticed that Richard Lindzen forecast that over the next 20 years, the climate is as likely to cool as warm, and said he would be prepared to bet on it. When Annan tried to arrange a specific bet, Lindzen insisted on 50:1 odds. At that rate, Annan didn’t see any upside to the bet. Annan’s characterization is that when Lindzen quotes odds that are that steeply in his favor, he is saying that he doesn’t have much confidence in his views. (I think he is precisely correct on this point. When I’m confident of a position I give my opponent odds so he is incented to take the bet.)

Anyway, this is the kind of conversation that I think Prediction Markets encourage. There’s a public debate going on about global warming, and in the popular press it’s hard to tell what each side actually believes and with what confidence. In order to have a bet, the two sides have to agree on what the terms are, and agree that they disagree on some specific prediction. They can disagree on the odds that some specified outcome will occur, or disagree on the value of some measurement that can be taken at a specified date in the future. In either case, there’s room for a bet. The best bets leave each side believing that they are getting better than even returns.

Annan makes this point repeatedly, and shows that he is looking for the kernel of the disagreement so he can create a bet that the other party will see as advantageous. He then points out that someone who gives odds for some event in a paper that will be referred to by the media and by politicians when making policy decisions they should be certain enough of the numbers to back them up with their own money, or there’s little sense in claiming that they represent the advocate’s best estimate.

Once a bet has been made, particularly if it’s done in a context like a prediction market where the odds represent the opinions of more than two antagonists, the odds and the prediction are publicly visible, and everyone gets a better idea of what the divergent views are.

AFAICT, all of the arguments made in the discussion about the value of betting were dealt with by Robin Hanson in his earliest papers on Idea Futures. Some examples:

• uninformed betters (The existence of profit potential will draw in more informed betters and give an incentive for contrarian research.)
• long time frames for results. (Robin suggested that the bet’s values could be stated in terms of a market-neutral instrument like T-Bills or the S&P 500. This would mean that the money held against the bet would appreciate like any other investment.)
• Manipulation (Robin has done recent experimental and theoretical work to show that manipulators increase the returns and incentives for betters to move the market in the correct direction.)

The latest cover story in BusinessWeek, The Power Of Us, is about using the Internet to harness the power of decentralized talent — from web services to let small merchants build big stores, to aggregating marketing information fragmented across engineers, salespeople, and customers using markets.

It also quoted two of the contributors to our recent workshop at Supernova 2005!

On Prediction Markets:

Eli Lilly & Co., Hewlett-Packard Co., and others are running “prediction markets” that extract collective wisdom from online crowds, which help gauge whether the government will approve a drug or how well a product will sell. … Corporate planners are even starting to use the wisdom of online crowds to predict the future, forecasting profits and sales more precisely. Prediction markets let people essentially buy shares in various forecasts, often with real money. Most famously, they’ve been employed in the University of Iowa’s experimental Iowa Electronic Markets to determine, with remarkable accuracy, the most likely winner of the Presidential election. The ease of organizing groups on the Net has caused an explosion in their use, says Emile Servan-Schreiber, CEO of NewsFutures Inc., a consultant that has run 40,000 prediction markets for companies and publications. Hewlett-Packard Co.’s services division was having trouble a few years ago with forecasts in the first month of a quarter. So Bernardo A. Huberman, director of HP Labs’ Information Dynamics Lab, set up a market with 15 finance people not normally involved in such planning. They bought and sold virtual stock that represented a range of forecasts at, above, and below the official company forecast. Their collective bets yielded a 50% improvement in operating-profit predictability over conventional forecasts by individual managers.

On Amazon Web Services (AWS):

At Amazon.com, thousands of volunteers write buyer’s guides and lists of favorite products. Amazon also lets thousands of merchants, from Target Stores to individuals, sell on Amazon pages. What’s more, Amazon is opening up the technology behind product databases, payment services, and more to 65,000 software developers. They’re creating new services, such as the ability to compare brick-and-mortar store prices with Amazon’s by scanning a bar code into a cell phone. Thanks in part to such moves, the company is solidly profitable on $6.9 billion in sales last year. “We’re all building this thing together — Amazon itself, outside developers, associates, and customers,” says Jeff Barr, Amazon’s Web services evangelist.

“… an economy of the people, by the people, for the people”

At the same time, peer power presents difficult challenges for anyone invested in the status quo. Corporations, those citadels of command-and-control, may be in for the biggest jolt. Increasingly, they will have to contend with ad hoc groups of customers who have the power to join forces online to get what they want. Indeed, customers are creating what they want themselves — designing their own software with colleagues, for instance, and declaring their opinions via blogs instead of waiting for newspapers to print their letters. “It’s the democratization of industry,” says C.K. Prahalad, a University of Michigan Stephen M. Ross School of Business professor and co-author of the 2004 book The Future of Competition: Co-Creating Unique Value with Customers. “We are seeing the emergence of an economy of the people, by the people, for the people.”

The Million-Monkey Money Manager:

One investment-management firm, Marketocracy Inc., even runs a sort of stock market rotisserie league for 70,000 virtual traders. It skims the cream of the best-performing portfolios to buy and sell real stocks for its $60 million mutual fund.

Recently, Sun has been running an intriguing double-page spread print campaign with various of their technical luminaries standing in a field, Hands-Across-America style. Some depictions are a bit disingenuous (Bill Joy), some are not as flattering as they could be, and some, indeed, are eyebrow-raising “Oh yeah, I guess so-and-so really is still there!”

This has intersected with Gary Anthe’s latest from-the-labs report in ComputerWorld. I believe he’s written several of the other lab profiles we’ve pointed to in the past. The following review is courtesy of ACM’s TechNews clipping service. In the extended entry, we’ve excerpted a bit on Vipul Gupta’s work with miniaturizing emedded SSL web servers using ECC…

Sun’s R&D Spectrum Computerworld (06/06/05) P. 29; Anthes, Gary H. Plus sidebar on supercomputing

Sun Microsystems employs some 200 scientists with more than $80 million to spend annually on a wide variety of next-generation computing projects, including a possible 4-PFLOP supercomputer and a Web server the size of a quarter. Sun’s Proximity I/O technology, for example, will enable computer chips to fully maximize their potential computing power so that top-tier Internet switches can be built at dimensions and costs similar to PCs; currently, Internet switches cost millions of dollars and fill entire rooms, but Proximity I/O eliminates wire interconnects and the data-transfer bottlenecks associated with them. “When processors went from 10MHz to 3GHz, they didn’t become 30 times faster because the bandwidth didn’t increase by 30 times; it increased by two or three times,” notes Sun Labs researcher Robert Drost. Sun leveraged the potential of Proximity I/O to win a DARPA bid to design and build the next generation of supercomputer architecture. Sun, IBM, and Cray won the three $50 million contracts, and one project will be chosen for actual production by 2009. Proximity I/O would enable massively parallel computation between large numbers of processors, lifting the sustained speeds of that machine above 1PFLOP, possibly scaling to 4PFLOPS. On the other end of the computing spectrum, Sun has developed secure, coin-size Web servers that could be deployed in battlefield sensors, on personal medical devices, or RFID tags used for confidential situations, and Sun’s elliptic-curve cryptography (ECC) is key to this effort because it dramatically reduces computing requirements compared to RSA cryptography while maintaining similar security.

(more…)

By now, most of you have seen announcements of the recent paper showing that Oxytocin increased the level of trusting behavior in an experimental setting. The aspect of these results that I want to focus on isn’t the medical implications, or the possibilities for abuse or for clinical use, but the fact that economics experiments are being accepted as the standard measuring tool for social behavior in a medical context. Three of the four authors of the paper are economists, judging from their affiliations. The results were published as a letter in Nature.

The economists used a couple of fairly standard game designs that measure trusting behavior in an interaction. The first of two experimental subjects can choose to trust the second, even though there are no later interactions that give the first player an ability to reward or punish the second player’s behavior.

The authors went to a lot of trouble to design an experiment (read the paper!) that distinguished between the trusting behavior of the first players and the trustworthiness of the second. Only the former was enhanced. They then add a variant in which the second role is played by a computer. The first players were more trusting of people under the influence of Oxytocin than without it, but their play when a computers was making the second decision was unaffected by Oxytocin.

The experimenters administered questionnaires to determine whether it was the subjects’ expectations about the outcomes rather than their level of trust that was affected by the Oxytocin. The subjects had indistinguishable expectations about how much they would get by trusting, but still the subjects trusted more under the influence of Oxytocin.

All the news reports I’ve read accept the experimenters terminology, and write as if they agree that what is being measured is indeed trust. This is the first case I can think of in which a medical intervention had effects on interpersonal behavior that matched the kinds of behavior that experimental economists have been probing for a while. I’m impressed by the level of acceptance of economic experiments as measures of something as evanescent as trust. That’s a level of respect I don’t remember seeing before.

Thanks to Ben Sittler for the initial pointer, and the hint that this was relevant to my interests.