Archive for December, 2008

Why GetSatisfaction is great

Monday, December 22nd, 2008

Suddenly even Fortune 500 firms like Apple are using GetSatisfaction, a sort of specialized social network for handling customer support.

GetSatisfaction transcends simple cost cutting.  Whether they know it or not, they are applying the “quality is free” ideas of W Edwards Deming to web-based support.  GS saves money and time for both vendor and user, while offering better support and lower cost, and actually improving the product at the same time.

How could it do all that?

To see, consider the challenges Eric and I faced with the products we created at Whitehorse Games in the 1990s.

Whitehorse was one of the early Web-based video game vendors, offering softgoods and Web-based software unlock in 1999.  This created new problems and opportunities:

  1. Customer feedback on a new release was instantaneous.
  2. Free downloads increased support requests by at least a factor of 10.
  3. 5% of problems created 95% of support requests, and you could discover this almost immediately.

Email load quickly became overwhelming:  users of demo software can be shockingly strident and demanding, considering that they have received something for nothing.  There is a tendency to just disengage:  forward the customer to a static FAQ that answers the most common problems, and deliberately delay personalized response in the hope they would go away.  This sounds harsh, but there are only so many hours in a day.  Whitehorse “solved” the problem by ending free trials — this reduced support load by over 90%, and did not hit sales at all, because, it turns out, most users of freebies will never pay for anything.

But I recall thinking, even then, that this missed a huge opportunity to improve the product.  Freebie users sometimes offer you something valuable when they take the time to complain.  If there were just some way to efficiently collect product suggestions, categorize them in a way that did not consume human time, the product cycle could be compressed to radically improve things in much shorter time.

In fact, in 2001, I briefly developed a startup idea around this, but my version was too complex (attempted to systematize the entire software process, including an outsourcing marketplace like Elance), too centralized (outsource support by direct-selling to individual vendors), and too narrowly marketed (attempted to deal only with software vendors).  

GetSatisfaction fixed all those problems and more.  They are a simple, broad solution to product support, with an efficient suggestion box built in.  Clever.  Hope they do well.

Here again, the goal.

Sunday, December 21st, 2008

$775 billion — the current proposed fiscal stimulus, aka deliberate deficit spending – is fearsome not for its scale, but for the risk it will be aimed at an unproductive goal.

Jobs are not the goal.  Increasing consumer spending is not the goal.  They are merely happy byproducts of something more important.

The goal — the objective of economic policy — is to permanently increase median GDP per work hour.

Every word of that definition matters.

Why per work hour?  Because no one is better off if GDP increases 10% by working 10% more hours, or by having a 10% larger population.  The measure of economic improvement must be GDP per work hour.

Why median?  Because if Wall Street earns an extra $100 billion, and everyone else stays the same, then GDP per capita went up, but in fact almost no one is better off.  Unlike the average, the median ignores statistical outliers — the best way to increase the median is to improve the productivity of the greatest number of people.

Why permanently?  Because otherwise, after the stimulus money is gone, you’re back where you started.  And anyway, simply handing out checks and keeping people busy is too unambitious. 

One cost effective way to permanently increase median GDP per work hour would be to spend stimulus money optimizing America for productivity and productivity growth.  Some easy examples:

  1. Move all government filings to Web forms, eliminating PDF and paper.  Concurrently, teach every adult to type and use a Web browser.
  2. Widely teach the benefits of productivity growth, so people know what solutions to look for.
  3. Widely teach probability and statistics, and their application to efficient production.
  4. Teach 10 million adult Americans to speak German, Japanese and Chinese.
  5. Translate all foreign production management publications into English, and republish them here.
  6. Eliminate the Big 3 automakers’ existing distribution model, and replace with a build-to-order model like that used for Mini Cooper.  Showroom contains just 1 or 2 of each basic model.  Fixed price.  You order the colors and options via Web form.  Car arrives a couple of weeks later.  This offers greater choice, while eliminating inventory, waste and the deadweight loss of the sales-push model.  Also provides instant product feedback from sales to manufacturer, allowing products to improve more quickly.
  7. Force the US military to use commercial platforms for its cars and trucks.  In particular, the Humvee would be replaced by an armored GM/Ford 4×4.  This would help GM/Ford, the Army, and the taxpayer at the same time.  For the Army, mass produced vehicles are not only cheaper, but also more reliable, cheaper to maintain, and better designed because each design sees more total use.  For GM/Ford, increased sales would push factory utilization back up toward breakeven, reducing average cost per vehicle.  For the taxpayer, the cost per Army vehicle would go down.  For the consumer, the price of 4×4’s goes down, because GM/Ford passes on some of its lower cost per vehicle.  Everyone wins.  It’s insane for the Army to use purpose-built designs.  They only do it out of momentum — it’s been that way since World War II.
  8. Change the tax structure to encourage production at the expense of consumption.
  9. Change the regulation of the media to encourage production-oriented content, and discourage consumption-oriented content.  (To see the pervasiveness of the problem, visit a magazine rack and notice that nearly every publication is essentially consumption-oriented.)
  10. Hire 5 times more patent examiners, double their salaries, and cut the small-entity patent filing fee.  Simultaneously, raise the bar on actually obtaining a software patent, particularly in software.
  11. Provide tax breaks for how-to publications that focus on exports and production management.

Instead of wasting money on yet more unused light rail lines — which are a nice idea, but wholly incompatible with existing city layouts –we could put $775b to real use.  Productivity-oriented education in particular has astonishingly high ROI compared with alternatives.

Freakonomics misses something

Sunday, December 21st, 2008

Have meant for years to write this.

Freakonomics (Levitt & Dubner, 2005) famously argued that Roe v Wade was the most statistically likely explanation for the dramatic fall in U.S. crime during the 1990s.  The hypothesis was that, after the early 1970s, unwanted babies were less often born, resulting in fewer maladapted folks to commit crimes 20 years later.

This was clever, unexpected and interesting, and did fit the data better than the half dozen alternative explanations the book considered.  However, it did not explain the dramatic rise in crime during the 1960s and 1970s.

Here is an alternative explanation the book did not consider:  post-traumatic stress disorder among returning veterans.  PTSD is well known to be linked to crime and violent behavior.  War-caused PTSD is known to trigger episodes of violent behavior, and over 30% of tested prison populations have PTSD.  

Large numbers of US soldiers began returning from Vietnam in the mid-1960s, about the same time as crime rates dramatically rose.  The rise in crime leveled off a few years after the US withdrew from Vietnam, and began to decline precipitously as the US veteran population aged.

Interestingly, the violent crime rate continued to fall until just a couple of years ago, when it turned back up again — coincidentally, just as large numbers of vets began to return from Iraq.

This idea leaves out a few things — for example, it does not explain why crime rates would not dramatically rise after WWII or Korea.  But for completeness, Levitt and Dubner should cover it.

America's Ace

Thursday, December 18th, 2008

Despite the debt, the profligacy, the failing basic institutions, America retains one unshakable core principle that constitutes a hard-to-copy competitive advantage:  the nation is defined by ideology, not ethnicity.

To see how big this advantage can be, consider the following solution to our industrial hollowing.  

For 20 years, we have received an unceremonious manufacturing beat-down from China.  But in the consumer-led recession, China is in perhaps worse shape than we.  they risk social instability unless 10% of GDP is generated from exports.  Sooner or later (probably sooner), central banks or no, the dollar will decline, and that trade imbalance will close.

At that point, assuming a dollar crash, it will suddenly be economically feasible to manufacture in the US, but because of the long-term hollowing, we lack the skill base.  

What if, at that point, America simply offered permanent residency to any person in China with a degree in mechanical engineering and 5 years experience designing manufacturing lines?

Instant result:  massive skills drain into the US.  Why?  For most people, it remains easier to start a business here, and easier to live an unfettered life here.

Could China do the reverse?  Not in a million years.  The nation is defined by ethnicity, so it cannot import talent.  That’s a weakness.

Rationalizing the auto industry

Monday, December 15th, 2008

In the political debate over how and whether to save GM and Chrysler, microeconomics has been forgotten.

Autos pretend to be a brand-driven industry, but are more like a commodity industry. Under conditions of overcapacity, the highest-cost producers of a commodity shut down, while the lowest-cost continue to make money.  These shutdowns are most efficient if they happen at the plant level, not the firm level.

It’s not clear that union-busting — currently demanded by the Senate minority in return for a rescue — is economically necessary or politically constructive.  Current GM workers are actually paid less than Toyota workers in the US.  GM’s higher costs result not from current staff, but retirement obligations, mainly retiree healthcare.

GM is in an interesting position:  they are uncompetitive mainly due to retiree benefits.  In fact, it barely exaggerates to say that GM is going broke due to the US healthcare system.  The system is not cost effective, and has become less so over time, to the point that it is beginning to bring down even very large firms.  Fixing the cost effectiveness of US healthcare would dramatically improve US competitiveness, profitability, employment — and could by itself save GM.

But that takes time.  In the meantime, if GM is to be saved, logically it should be done as a workout or prepackaged bankruptcy, in which retirement liabilities are nationalized and the oldest, highest-cost plants are shut down.  GM is competitive enough that, with these two changes, they should survive.  Without these changes, they’ll just fail again.