Why We Overestimate the Short Term and Underestimate the Long Term in One Graph

We always overestimate the change that will occur in the next two years and underestimate the change that will occur in the next ten.

- Bill Gates, The Road Ahead
I'm always suspicious of aphorisms. Usually you run into this situation. But Bill Gates' often-quoted line above is one of the better ones I think. I do believe, though, you could add to it -- i.e. that in 20/30/50 (?) years, our hopes and expectations about the future often tend to converge with reality (e.g. ARPANET -> today's Internet).
I got to thinking why Gates' rule of thumb seems to consistently hold true. It feels to me it basically boils down to two curves graphed over one another:
Of course the above is an over-simplification, but as you look at the curves, you might start to surmise a few things:
  • Carlota Perez and others have characterized the S-curve (a sigmoid function to be precise) as an installation stage leading to a deployment stage. To me it looks like three main phases that could be characterized as a) a slow build up to some kind of critical mass, b) compounding taking effect and driving exponential growth, c) ultimately diminishing returns once some kind of limiting factor (e.g. number of hours in a day we're awake, number of humans on Earth, etc.) starts to take effect. 
  • If the intersection point is truly somewhere between two and ten years as Gates suggests, those five year strategic planning processes that stodgy corporatations love so much may not be such a bad idea in tech after all.
  • Maybe there's an evolutionary reason why we tend to think linearly. But maybe evolving to thinking more like an S will be one of the signs of the Singularity :)

Proposal: New Conditions for Patentability

Patents are a hot topic these days. A lot of the debate tends to center around notions of fairness or finding ways to apply current standards better, which I sense hasn't resulted in much progress. Some do argue for fundamentally new ways to define what should and shouldn't be patentable, but what I've seen so far strikes me as either a bit narrow in scope or too sweeping. I think there's something in the middle to consider. More on that in a sec.

First, let's revisit the traditional argument for patents to exist in the first place -- i.e. inventors will be less motivated to bring their creations to the world if others can easily copy their idea. Therefore, giving them patent protection should result in a boost in innovation, economic productivity and general consumer utility.

A BU School of Law paper from a few years back attempted to gather empirical evidence on the premise above employing the same methodology used to demonstrate the positive economic impact that most economists agree are associated with basic property rights. The findings if anything support the converse:

 ...the empirical economic evidence strongly rejects simplistic arguments that patents universally spur innovation and economic growth. The direct comparison of estimated net incentives suggests that for public firms in most industries today, patents may actually discourage investment in innovation. 

As with any study, we should exercise a healthy level of skepticism. The authors, Bessen and Meurer, are affiliated with an organization that may or may not have a specific agenda, but to date I've seen only respect for their work, even among critics and there are many others that have come to the same basic conclusions. (Plus, I kind of like the fact that a major law school published a paper that in effect makes a case for less lawyering.)

Now, even if we take as fact that patents generally fail to spur growth and innovation, it's probably unwise to look to blanket abolishment. The authors are careful to note there are exceptions, e.g. in the pharmaceutical industry, where you have exceedingly high upfront R&D requirements and long development cycles. Additionally, there may be cases where downside risk is just too high to remove patents completely should it disrupt an entrenched system of creation and delivery for products and services that are critical to human well-being (again pharma is a pretty good example).

Instead, I think the current definition and conditions for patentability need to be expanded to establish a higher standard. In addition to the general subject matter described as patentable, and the conditions of usefulnessnovelty and non-obviousness, I'd add at least one more: 

  • High Barriers to Entry: The problem of 'free-riders' is most acute when the originators invest large amounts of money and time upfront only to let others copy them without having to bear the burden of those fixed costs. This is a legitimate concern and I believe a valid and practicable criterion for assessing patent-worthiness. The PTO could use direct data or industry benchmarks for average R&D capex and average time to market to serve as reasonable guidelines. In the process, they'd get to filter out a huge amount of their infamous backlog, focusing on those applicants that are taking the most economic risk, and not getting overwhelmed with trying to learn intricacies of new technologies that are hard enough to keep up with at a cursory level.

 and possibly:

  • Essential Societal Value: Admittedly this is harder to quantify, enforce and de-politicize, but could at least be a factor in prioritizing application reviews if it isn't already. Ultimately, patents are supposed to provide benefit to society by ensuring the populace doesn't get deprived of important creations. Some sort of most favored nation status for industries like healthcare/pharma, energy, defense, critical infrastructure, etc. may be warranted. The truth is, some inventions are just more important than others


In closing, a quote I saw recently on Chris Dixon's blog:

That ideas should freely spread from one to another over the globe, for the moral and mutual instruction of man, and improvement of his condition, seems to have been peculiarly and benevolently designed by nature, when she made them, like fire, expansible over all space, without lessening their density in any point, and like the air in which we breathe, move, and have our physical being, incapable of confinement or exclusive appropriation. Inventions then cannot, in nature, be a subject of property.

         - Letter from Thomas Jefferson to Isaac McPherson

This from a Founding Father, but also a bit of a tinkerer in his own right.


Keynesian Schumpeterianism

Albert Wenger penned a pithy line a couple months ago about the possibility of "more governance (broader role) while at the same time having “less government” (fewer dollars)." It spurred me to dust off some of my own thinking on the role of government in propelling our economy. It stems from a simplistic understanding, cherry-picking and eventually melding of two legendary economists' theories, traditionally considered at odds with another:

Among champions of innovation and entrepreneurship, the gentleman on the left, Joseph Schumpeter, is a hero. He made the term "creative destruction" popular and is often credited with being one of the first to apply economic theory to entrepreneurship, and vice versa. In this context Schumpeter's thinking can be boiled down to: 

  • Dramatic leaps in innovation are required to sustainably drive productivity, which is the root force behind economic growth.
  • Disproportionate, short-term rewards are necessary to provide incentive for risk-takers to pursue such innovation.

On the other hand, John Maynard Keynes, the mustachioed fellow on the Time cover, is frequently associated with the notion of active fiscal management to mitigate downsides of business cycles. Keynesians traditionally believe:

  • All-out free markets tend to cause a lot of pain (i.e. extended periodso of high unemployment) in the real world.
  • You can manage these downturns by saving on sunny days and spending on rainy days, even with money you don't have. 

As you can imagine, pro-business types and entrepreneurs tend to take more to Schumpeter. And how can you blame them – the guy allegedly wore a cape on a regular basis, and clearly was a bit of a badass (for an economist):

Schumpeter claimed that he had set himself three goals in life: to be the greatest economist in the world, to be the best horseman in all of Austria and the greatest lover in all of Vienna. He said he had reached two of his goals, but he never said which two although he is reported to have said that there were too many fine horseman in Austria for him to succeed in all his aspirations.

Let a playa play!

Keynes by comparison was "the guy trying to turn the party down just as Schumpeter was coming in from the kitchen with another bottle of tequila."

<Sad trombone>

As our government is making decisions on economic and budgetary policy and people continue to debate “Schumpeter or Keynes?” (or some equivalent)…I think the answer is a resounding “Yes.” It’s seductive to go all in on the idealism and theory of the former. The reality is that we as humans are subject to non-rational, non-efficient behaviors and outcomes. This in turn leads to 'non-linear strife' and possibly even societal unrest during extended periods of suffering, so pragmatically the two schools of thought must go hand in hand. At least until Ray Kurzweil is proven right.

If you think about it graphically, Schumpeterians focus on uninhibited innovation – even at the cost of periodic destruction – driving sustainable economic growth [sine wave with upward slope], while Keynesians are concerned with minimizing volatility [lower amplitude, lower frequency]. Putting the two together, one can see they need not be mutually exclusive and that it’s not unreasonable for government to explicitly set and manage to economic goals based on both ideals: 

With all the political bickering and lingering problems with the economy, I'm hoping policy makers and voters can appreciate a more holistic view, and that largely their arguments are two sides of the same coin...(or wave, as it were). Maybe to some that's been clear all along, but I sense that most in this country are exposed only to talking points from one extreme or the other. The truth is, even Keynes and Schumpeter themselves it turns out had a genuine, if grudging, admiration of one another:

Schumpeter...told his students...that Keynes's work had totally superseded his own earlier writings on money...Keynes, in turn, considered Schumpeter one of the few contemporary economists worthy of his respect.



Four Flavors of Freemium

The freemium business model has become pretty pervasive ever since the term was coined on Fred Wilson's blog. There’s even a Freemium Summit now, and just made some waves announcing their first real freemium product.

But despite the rise in prominence, I haven't seen a lot of good frameworks for determining how freemium should be executed; mostly I see isolated case studies. I thought I’d give a crack at something a little broader. I’ll assume we’re talking about 'true freemium' here – i.e. includes an offering that is useful to a substantial set of users and is indefinitely free. Free trials (trymium?) don’t count. I’ll also leave out discussion of ad-supported approaches for now.


Resource Efficiency

First off, I think any business considering a freemium model needs to be able to deliver its core product with a lean resource base, at least during initial stages. If this isn’t true, the freemium option is probably a non-starter. After all, you’re counting on maybe only a percent or two of your entire user base ever paying you a dime. I see companies like Ning abandoning their free products, and then I see how much capital they’ve raised that they’re expected to make a return on, but before the fundamental freemium model has been proven. For freemium in particular, I think that's a backwards approach to scaling (see lean vs. fat startup debate for more).


After ticking the efficiency box, there are two dimensions I focus on:  

Product Enhancement

This is the obvious one – most profit-maximizing businesses recognize they have to offer more to paying customers vs. those using the free product. The key is deciding where to draw those lines. This can be done using tactics like cohort analysis, A/B testing, price sensitivity analysis and general funnel management. You can get very scientific about it, and there's probably an art to it too, but for those singularly focused on converting free to paid, you at least you have a pretty clear objective. In cases, however, where scale in the overall user base is important as well, you may need to be a bit more strategic.  More on that later.

User Interdependence

This is the component that I’m consistently surprised is overlooked, or at least not given its due. When you’re giving away your product to so many users for free, in many cases there are interdependencies that can emerge among them. Some may be more natural than others - e.g. for a company like LinkedIn or Skype, the user-to-user connections are pretty self-evident. But even taking a service with a seemingly standalone customer it’s often possible to draw interdependencies out to produce some base level network effect; i.e. Metcalfe's Law (although probably not to an n2 degree). At the very least, this will make your product less vulnerable to substitutes since it'll be less attractive to switch.

Let's take TurboTax as an example. You can use their free service if you have low gross income or a basic filing or you can pay for more complicated filings, additional state returns and/or ancillary services. This could easily be siloed for a given user, but Intuit realizes there’s value to showing how your tax return compares to the rest of their user base, as well as providing online forums for advice on money management from their community. Even if these aren’t primary motivators in the initial purchase process, they at least help with customer satisfaction and retention. Competitor X may have a couple features that are better, but they can't give me the benefits of accessing a large network of other users.

Four Flavors

Combining the two dimensions above, we can organize them into a tidy 2x2 matrix that any MBA or management consultant could love (I was guilty of both), from which four freemium models can be described: 

  • The Tip Jar Approach:  Here, everyone gets pretty much the same product. Some throw you a bone out of a sense of charity or obligation. This is how street performers earn a living, and basically what Radiohead’s famous In Rainbows experiment was about (although they eventually ended it, which is probably telling). "Donate" buttons on some sites and services across the web are also examples of this.
  • The Special Member Model:  Everyone gets the same core product here too, but there's an added benefit that users get from the community aspects of the product. A church is an age-old example. NPR stations are another – I can listen to KCRW for free all I want, but I can also pay a little something and get a tote bag to show others I support public radio and get invited to exclusive events with other progressive, hip, pretentious members like me ;)
  • The Standard Upgrade Model:  I think this is what most people think of when they hear “freemium.”  I pay to get more stuff than the free product has.  Pretty simple. But that’s really all the additional value I get. Evernote and HootSuite – both great services – would fall in this camp as would many others. It's certainly possible to build a solid business with this approach, especially if marginal costs and revenues are aligned and conversions are executed well, but I’d argue there’s a missed opportunity here, namely…
  • The Network-Upgrade Model:  Ah yes. Up and to the right - where the magic happens. This is where you’re not only providing more value to the user in enhanced product attributes, but you’re also providing them more value from the interdependencies that exist from the rest of the user network. In addition to LinkedIn and Skype, more contemporary examples of companies adopting this model include Dropbox and Disqus (disclosure: I work for Disqus so consider me heavily biased). Dropbox and Disqus are particularly apt examples given their comparatively lean resource bases. 

For-Profit vs. Not-for-Profit (y-axis)

Looking along the vertical axis, it’s apparent that the non-differentiated product is probably not viable for most profit-seeking enterprises. The examples on the bottom two quadrants are pretty much all not-for-profit (including the Radiohead example I’d argue), and often have to rely on subsidies from other sources (e.g. grants for NPR, being rich and bored for Thom Yorke). Not that surprising really.

Marketing vs. Marketing AND Product (x-axis)

Moving across the horizontal axis, though, it’s more interesting. You realize that those to the left are mostly using the free product as a marketing vehicle, whereas those on the right get to leverage the free product as both a marketing/conversion vehicle and a major component of their core product’s value to the user. i.e. The free Dropbox service not only helps me enter their conversion funnel, it also makes Dropbox way more valuable to me since a lot more people, some of whom I’m going to want to transfer files to or from, are going to use Dropbox. Similarly, the core Disqus service for publishers to enhance their comment discussions is free and might lead me as a blogger to purchase one of their premium add-on packages one day. But even if I never do, I've now done my small part to make the Disqus network that much more expansive and have increased the federation and cross-pollination benefits for all other website communities using Disqus, whether TechCrunch or some random Tumblr blog like this one (wtf?).

In summary - if it’s not already obvious - I’m a big proponent of that upper-right, ‘Network Upgrade’ box. I think it’s a missed opportunity to not move in that direction for so many freemium businesses out there that may not think to build in those dynamics. And per the TurboTax example, even if it's not a primary consideration, I don’t think you have to be a purely intrinsic network to produce the sorts of interdependencies we're talking about.


Bonus: Freemium Networks - Head and Tail, Go for Scale; In the Middle, Monetize More Than a Little <sorry, I desperately wanted that to rhyme>    

So if you buy into all this, a logical question is: How do I balance the sometimes competing priorities of scaling to produce network effects vs. making billz?? (I mentioned I'd get back to this in the Product Enhancement section above.)

Well, here’s one thought on that, especially for cloud/software as service. On average, your smallest users presumably have the least to spend but may well comprise a large base of users, particularly given the theory of long tails. Therefore, focus your efforts on scaling on that part of the curve like a mofo. Make sure your free product serves that segment really well. 

On the other end of the curve are the big fish – they probably have some money to spend, but they are limited in number, may be slower to move and most importantly, may be as or more valuable as vehicles for scaling. I wouldn’t necessarily give away your product to them – chances are they are going to want more than your free version provides, and plus they tend to be suspicious of free lunches. But in this scenario, it may behoove you to price at or near cost as opposed to trying to make a big margin from them. They have lots of strategic value in terms of expanding your user network and/or lending brand credibility. 

Finally, there’s that juicy middle part. This is potentially your sweet spot where you have prospective users who have money to spend and are relatively plentiful, but it’s not the end of the world if they don’t end up using you because they don’t account for that much scale in aggregate. Plus, they’re probably less expensive to serve relative to the big guys in terms of upfront sales effort and hand-holding. 

This is of course theoretical and somewhat simplistic. There will be exceptions and overlaps for sure. But if you are a believer in the Network Upgrade model, think about the monetization vs. scaling balance in this way. It may make sense for the long-term value of your freemium business.

And above all, think about how you can use free as a way to make Metcalfe's Law work for you, not solely as a means to convert to paid (but don't slouch on that either).


Thanks to Ki Mae Heussner, Raghav GuptaSachin Agarwal and Daniel Ha for reading drafts of this, and to Paul Graham for giving me the idea to do this thanking bit so I look like a mentsch.

This article was also published on Business Insider -- comments may be viewed and submitted to the same discussion thread there as well.


Hello, Again, World

It's been a few years since I've regularly maintained a blog, and even then it was really just a travel blog I kept up while living overseas for a few months. 

I'm hoping to start up again and make this a place to semi-regularly post thoughts on whatever it is around me that seemingly warrants a reaction from yours truly. At the moment, that happens to be a lot around things like business strategy, product development, economics and general philosophy, especially as those topics relate to startups and the internet given how I spend my time these days.  

This may change in time or I may deviate once in a while with something on a band I like, or a team I root for or a movie I'm trying to figure out .. but whatever my thoughts happen to be on any given day I try to approach everything with two lenses:

1. Logic  2. Empathy


 ...usually in that order.  I find that as long as I tick those two boxes, I'm usually pretty satisfied with the result.

Hope you'll agree, but don't be shy to let me know either way.