Continuations

There is a lot in the FY 2010 US budget and it will take some time to digest.  But one thing that is clear right away is that there is a big tax hike coming for folks at higher income levels.  I for one have never been upset about paying taxes.  That’s probably in part because growing up in Germany my Dad, who was a teacher and — like all teachers in Germany at the time — a government employee, instilled a sense that paying taxes was a civic duty right up there with voting.  I still believe that today, although it is a lot harder during times when a lot of money is spent by government in ways that I disagree with (this certainly applied to the previous administration, but also to big parts of the bailout of the banks).  I wish though that much of the tax increase would be achieved by radically simplifying the tax code and eliminating loop holes that let some folks pay virtually no taxes even at very high income levels (ditto for corporations).  Instead, the rules seem to limit mortgage and charitable deductions, both of which would seem to have side effects counter to what we need right now.  Some interesting weekend reading ahead.

Upfront: I know much has been blogged (and tweeted) about this recently but I am writing this on the train so no links - sorry! I am on my way to another meeting about the economic future of the city. These meetings have resulted in a series of initiatives announced by Mayor Bloomberg last week. There has been a fair bit of commentary on the substance already, but I would like to address the process. I believe it could be hugely improved by opening it up. For starters, the folks running this should have a web site, which could of course simply be a blog, that talks about the goals for the project, who is participating, how the city is going about this and so on. That site should be open and allow comments. Sure, there will be some useless chatter and possibly some flames or worse, but that seems like a small price to pay for allowing anyone to contribute. Individual meetings like the one I am headed to should be open too. By that I don’t mean anyone can show up - there is definitely room for invitation only discussions in which a smaller group can talk about a subject, but I see no reason why those meetings should not be bloggable and in fact be blogged both on the site I suggested above and on the sites of participants. I believe everyone is genuinely trying to help and would continue to express their opinions whether or not someone else can write about them. In any case, I am fairly suspect of any view on an important public matter that someone is willing to express only in a private non-bloggable setting. Any work products should also be public. That includes presentations created by consulting firms, but also draft action plans, legislation, etc. Those documents should be under revision control so that one can track their evolution over time. I firmly believe that the benefits of such an approach vastly outweigh some of the extra noise that it will generate. The old saying was that ‘politics is like sausage - you don’t want to know how it’s made’ - but we know where that view has gotten us. Now it’s time transparency and engagement.

When we invested in 10gen last year, the team was working on a very ambitious approach to deliver an integrated cloud platform a la Google App Engine.  While they made amazing progress on that, it turned out that developer interest was focused around the MongoDB datastore.  Instead of having to buy into a complete stack, most folks (at least for now) want to continue using a broad choice of technologies and combine elements as they see fit.  This appears to be one of the key factors in the tremendous continued uptake of Amazon’s EC2 and AWS offerings compared to Google App Engine.  The 10gen team therefore decided to focus all their efforts on the MongoDB datastore and I am excited that they delivered an initial standalone release earlier this month.

Who should try MongoDB? Everybody building new web sites or services or operating existing ones.  Mongo is an operational datastore, which means that it is aimed at the data that powers the interactions with the site or service (as opposed to say analytics).  The team has a more detailed description of the possible use cases for MongoDB and also sets out some areas that it is less well suited for.

Why should you try out MongoDB? Because MongoDB is a much better fit for most web development than a traditional relational database.  Instead of requiring an ORM layer, MongoDB simply stores objects as documents in the database.  This is very fast since it eliminates a lot of overhead and therefore scales much better than a relational DB+ORM.  Yet it retains all the flexibility for super agile development.  Need a new field in your objects?  Just start saving new objects with that field.  Need a new collection of objects?  Just start saving to it!

You could get that kind of performance and flexibility with a key-value store (either on disk or in memory), but you would be sacrificing a huge amount of functionality.  MongoDB by contrast has rich commands such as retrieving only those objects from a collection that meet given criteria and then returning them in sorted order.  The kind of thing you will need in almost any setting (e.g., returning a paged list of results) and would not want to spend time handcrafting.

What is there today? For a first release MongoDB comes with a great set of capabilities, tools and drivers.  On the capabilities side, especially worth mentioning are, full index support, query profiling and master slave replication.  On the tools side, the command shell is totally awesome.  In it you can write things like

db.people.find({name:”albert”}).sort({age:+1}).limit(1)

which shows off just some of the shell’s expressiveness (there are also import and export tools).  As for drivers, there are already C++, Java, Python, Ruby and PHP drivers and there is even a community-contributed Erlang driver.   I should also point out that while the documentation is still growing, you can already find a lot of highly readable documentation.

What is next? I am extremely excited about the innovation push in database technology that is taking place.  We have had an amazing flourishing of new languages and frameworks over the last decade (although many of those efforts go back further), but underneath it our progress on persistence has been much slower.  Now with MongoDB, Couch, Hadoop, Hbase and many others.

The 10gen team has set out a roadmap for the development of MongoDB through the rest of the year.  They are looking for folks to experiment with this release and welcome any and all feedback on what’s there today and the plans going forward.  So start downloading!

We don’t  do very well with predicting the future because we fail to anticipate non-linearity.  This is apparent almost anywhere you look.  Most people when they hear that the average temperature of the earth might rise by a couple of degrees, seem to have a mental image of a slightly warmer summer and maybe a bit less snow in winter (if you live in a place that has snow in winter).  Very few people associate such a seemingly small change with the possibility of deserts in places that are lush today.  Yet with non-linear systems that’s exactly what you can get from a small change.

Here are some other recent examples that I have run across.  Apparently the plane that crashed in Buffalo was on auto-pilot during much of the time that ice built up on the wings.  Aerodynamic lift is a perfect example of highly non-linear system.  If you lose attachment of the airflow to the wing due to excessive icing or due to too high an angle of attack, lift does not decrease gradually, it simply disappears.  The easiest way to think about this is by imagining a glass on a table.  As you push the glass closer to the edge of the table, the glass stays at the same “altitude.”  That’s even true if part of the glass is already hanging over the edge.  But push just a tiny bit further and the glass drops to the floor.   Even in this seemingly trivial example it appears that we don’t have a “built-in” safety model, as kids will inevitably put plates and cups down at the edge of a table and learn not do so only after considerable breakage.

The current financial crisis is of course another massive case of non-linearity.  I have posted previously about how leverage vastly amplifies risk in a completely non-linear fashion.  The overall financial system is full of feedback loops and most of them are positive loops, meaning that effects are amplified resulting in non-linearity.   For instance, as the stock market drops people are less wealthy.  When they are less wealthy their tolerance for risk goes down and many stocks that previously seemed acceptable no appear too risky resulting in more sales.  That of course drives the stock market down further.

Yet another great example is server load.  We were talking to one of our portfolio companies yesterday that is experiencing rapid growth.  One of the founders observed that their database server has fairly low load and even during spikes does reasonably well.  He seemed to infer from that that they could handle much higher loads.  But such an inference is deeply flawed and ignores the many fundamental non-linearities in server load.  Let’s use a coffee shop as a simple example.  As long as folks arrive at a rate that is less than the rate at which the barristas can make espressos, lattes, cappuccinos, etc there will be no build up of a line (for simplicity I am assuming a deterministic coffee shop, i.e. folks arrive at exactly the same interval and it takes exactly the same time to serve them).  The second the arrival rate of customers exceeds the service rate, a line will start to form and that line will continue to grow as long as the arrival rate exceeds the service rate.  So a tiny change in arrival rate will result in a huge change in wait time!  So just because your server can handle current spikes does not mean it won’t completely croak on spikes at only a slightly higher level.

I have been reading a lot about education recently and how poorly it prepares us for a rapidly changing world.  I was looking for examples of that from my own life and it struck me how little I learned about non-linearity in school or even in college.  Given our apparent lack of built-in understanding of non-linearity this is a huge omission!

One of the great privileges of being in venture capital is the ability to observe many CEOs in action — not just at a distance, but close up.  One of the lessons I have learned from that is that there are different ways to be effective as a CEO and that very different personalities can make for effective CEOs.  I have observed effective CEOs who are technologists and others who are top sales people; CEOs who are methodical and others who are inspirational.  You get the idea.  But all the effective CEOs have one thing in common: they do what they do best and surround themselves with great people for everything else.

That of course sounds a lot easier than it is.  For starters, it requires knowing what you are good at (which does not have to be a function like engineering or marketing but could be cross-functional such as strategy or be a people skill).  Even that is not easy because too often CEOs seem to think that they have to be good at something other than what they are actually good at because they have mistakenly convinced themselves that something is critical to being a CEO (or worse yet because they believe that their board or investors expect it of them).  So to be effective as a CEO you have to be honest at least with yourself and ideally with your board about your strengths and weaknesses.

Then there is surrounding oneself with great people.  There are many reasons why this is hard.  But it is virtually impossible if the CEO does not see this as essential.  As long as there are one or more major positions or functions in a company that are not filled with or run by great people, it will be a drag on the effectiveness of everybody else.  Back to the first point though, seeing building the team as essential is different from doing it oneself!  I have seen effective CEOs who personally recruited every key member of their team and I have seen others who worked successfully with internal or even external recruiters to do so.

The great variety of styles and personalities that can make for effective CEOs is one of the reasons why I believe it is a mistake to early on look at a company founder (especially a first time founder) and conclude that they need to be replaced as CEO.  That should only be necessary if they either don’t know what they are good at (and by extension what not) or how to surround themselves with great people (even with the help of the board and/or recruiters).

I was excited about Obama from before the primaries got started.  Now his administration is getting off to a fairly rough start and it will be a true test of the man to see how he deals with it.  It seems pretty unfair to me that some folks are already loudly proclaiming their disappointment.  Given what’s going on in the economy (and the world), it seems unrealistic to me to expect everything to go smoothly.  So I for one am willing to wait a bit to see what kind of course corrections take place.

I try to do the same as a board member.  Stuff does go wrong in startups all the time, even with the best preparation.  The question is what happens after it does.  If everyone goes along with business as usual and nothing changes, that’s when you have a real problem.  But if there are lessons learned, the course corrected and — if necessary — changes made in the team, then you often wind up with a stronger and better company than if nothing had gone wrong.

As a board member I can make suggestions for what to do.  For the adminstration I can and will blog thoughts here (and send Facebook messages) but in the end just have to sit back and watch.  So here are three areas where I will be looking for change (so to speak): the Geithner “plan” for fixing the banking system, the contents and actual implementation of the stimulus, and the handling of the State-Secrets Doctrine.

Yesterday, the New York Times ran a story that described the paper’s strategy as trying to be the Last Man Standing.  Nicholas Carr today picked up on that in his post on newspapers today and writes:

Make sure you have enough cash to ride out the storm, trim your spending, defend your quality and your brand, expand into the new kinds of products and services that the web makes possible and that serve to expand your reader base. And then sit tight and wait for your weaker competitors to fail.

He goes on to say

Once you radically reduce supply in the industry, the demand picture changes radically as well. Ad inventory goes down, and at rates go up. And things that seem unthinkable now - online subscription fees - suddenly become feasible.

This argument is based on the idea that the removal of geographic boundaries has created a vast oversupply of news relative to demand.  While that is certainly true and while it is an important part of the problem, there is another part that this analysis completely ignores: The unbundling of content.

In the age of a physical, expensive to print and distribute newspaper it made perfect sense to get political, economic, culture, sports etc coverage all from one source.  It would have been cost prohibitive to get it from many different sources!  But this was an entirely cost induced form of aggregation.

With the web, however, the point of aggregation is shifting to the reader, with the help of intermediary services such as Google Reader, Techmeme and many others being built.  I can get my sports coverage from say ESPN, my politics from say Politico, my economics from say a variety of blogs such as Becker-Posner, and so on.  In fact, I wrote a post a while ago where I titled “Who Will Be My New York Times?” where I talk specifically about this.

Therefore I don’t believe in the Last Man Standing strategy for a broad paper that covers the whole gamut.  That kind of supply based bundling does not make sense going forward.

I do, however, agree with two parts of Carr’s piece.  First, micropayments as a salvation as proposed time and again (most recently by Walter Isaacson) won’t work for the reasons he lays out.  Second, I believe that there is a future for subscription services but it is much more likely to be for deeply focused offerings.  Imagine someone who is an expert on biotech.  They could use a free blog to build an audience and reputation (and SEO juice) and then offer a paid subscription with more in-depth analysis or with a timing advantage on a subscription basis.  Such a business could be extremely profitable because the focus will help keep the cost low even if this person were to employ analysts to help gather data.

Today our portfolio company Clickable has released version 2.0 of their Pro Product.  There are many great features in it, including support for emerging Google ad formats, keyword filtering & bulk edits, and enhancements to the Act Engine.

What I am most excited about though is that Clickable’s conversion tracking is now out of Beta and generally available:

Clickable Conversion Tracking is a powerful, independent, performance analytics tool that works across the three major search engines with one pixel.  It simplifies installation, automatically tags your destination URLs and gives you richer more consistent way to measure advertising return. It is an integrated, intelligent solution, as conversion data is used to set custom goals for your accounts and campaigns and it also informs Clickable’s ActEngine in real time, delivering recommendations customized to your goals and performance with greater returns.

This is a crucial step on the path towards letting small to mid-size businesses do what large enterprises have been doing for a while: picking the best advertising based on actual measurement of effectiveness.  This was one of our original motivations for the investment in Clickable, and I am happy to see the team delivering on this vision.

We are spending a lot of time thinking about how identity will play itself out on the Internet.  We are constantly running into the limitations of the existing arrangements even when companies from our portfolio are trying to collaborate.  While there is definitely movement afoot with Facebook, Google and others extending their authentication to third parties and possibly moving the OpenID standard along (see previous posts on this).

But there is also another candidate for identity at least in some situations and that is the mobile phone number.  I was reminded of that several times yesterday.  First, I met with someone who has been living in San Francisco for quite some time but still has her 917 cell number.  That made me realize that I have had my cell number for over 10 years and can’t imagine changing it voluntarily going forward.  Then I spent some time with Jeff Lawson from Twilio, which makes it super easy for web developers to add voice interaction to their services.  We talked about how IVR is often a pain, but generally that’s the case because the call starts out knowing nothing about you.  Since mobile phone numbers change so rarely, that does not have to be the case!  When I call say an airline, it should know who I am and immediately offer information directly relevant to me, such as whether my flight is on time.  A good example (surprisingly) is the New York Times, which when I call from my home phone pulls up all my information and makes reordering a missed weekend delivery a cinch.

I know that phone numbers can be spoofed via IP telephony hacks, so I am not suggesting that the mobile phone number can easily be turned into a reliable form of identity for security critical applications, but it could be used much more extensively than it is today.  This is especially true when you look at some type of multi-modal integration, such as calling in, being recognized and then being able to receive information back via SMS.

The current approach to the banking bailout is distinctly WOBW — worst of both worlds.  Somehow nationalization has been deemed not an option.  So we keep running bankrupt banks as private enterprises governed by existing equity holders with all the wrong incentives (worst of private).  That in turn necessitates having to regulate operational details such as compensation and lending practices with crude across the board measures such as $500K salary cap (worst of public).

The only way to get incentives properly aligned here is to wipe out the existing equity holders, set up an interim governance and create new proper incentives from scratch.  This would allow us to accomplish the crucial bits:

1. We don’t need to artificially determine a price right now for bad assets.  We simply get to keep the bad assets in a pile that we can slowly work off over time.  In any of the currently contemplated regimes (including the sidecar one proposed by Soros), we have to come up with valuations right now for the bad assets.  There is no way to get those numbers right (there are no right numbers!) and getting them wrong will either keep the firms bankrupt or will transfer wealth from taxpayers to equity holders.
2. We can immediately free up the good assets by spinning out parts of banks (and insurers) that are in ok or even good shape.  The management of these new entities can have significant equity (vesting over 4+ years) in lieu of exorbitant cash comp.  This would not have to be regulated — the S1s for these entities would detail the compensation of the management teams.
3. We can reshape the industry structure by creating many smaller stand alone entities which will make the industry more resilient and will make it much harder to hide risks inside of some arcane part of a huge firm.

If we are so hung up about “nationalization” let’s call it “supervised restructuring” but let’s get on with it instead of continuing to combine the worst of both worlds.

Yesterday’s piece by Claire Cain Miller in the NYT titled Angels Flee From Tech Start-Ups prompted many blog responses, including one by Roger which also got picked up by Bijan.

Here is what it Roger wrote on his blog

Are fewer companies getting funded by angels? Of course. Should all the companies that had previously been funded by angels, many of them dabblers, gotten funded? Definitely not. So is value creation in the early-stage space really being inhibited by a dearth of angel investment? I don’t think so.

Bijan more or less agrees with Roger that the good deals are getting done.

While that may well be true, I believe this view misses a crucial point on seed financing.  If only the good deals are getting done, then not enough deals are getting done!  That’s for two primary reasons:  First, the social value created by a startup (e.g., experience for the founders) far exceeds the benefit that accrues to investors.  Second, among even the most unlikely tier of startups there will be some real winners.

Dabblers therefore play an important role in the overall system.  Case in point: Etsy!  The original seed funding for Etsy did not come from and I believe would not have come from any of the professional angels operating today.  When Etsy got going, none of the founders had any entrepreneurial track record to speak of or built an ecommerce site (or any high traffic site for that matter).  Add to that the scepticism that many professionals (not just angels) had about the size of the market for handmade goods.  The first money for Etsy came from two Brooklyn real estate guys who had never before made a tech investment.

With the current retreat of dabblers, the next Etsy may well not be funded and that would be a shame.

Bijan has a post relating his recent overnight hospital stay to some of the big issues with healthcare today.  Bijan rightly complains about how hospitals are a black hole for patient data.  The stimulus package (PDF) contains a section 9202 titled “Investment in Health Information Technology” which states:

The Secretary of Health and Human Services shall invest in the infrastructure necessary to allow for and promote the electronic exchange and use of health information for each individual in the United States.

This is laudable, but as I have written in a previous post, as important as investment are changes in the regulatory framework.  This is very much the case for Healthcare, which is full of gatekeepers with entrenched interests, including not just the much maligned insurers, but also healthcare provider organizations (in what feels like a lifetime ago I had a startup selling to hospitals and integrated delivery networks).

So with respect to electronic medical records, I believe that we need a two pronged regulatory approach.  First, patients need be given electronic access to their own medical data.  This has to be made a requirement for all participants in the medical system by some reasonable date.  The government should invest it’s money in defining a standard for access that is easy to implement.  So no attempt to standardize vocabulary or anything else that would bog things down for another twenty years.  Simply an access protocol.

Second, there should be some requirements established for the operation of patient centric electronic medical record service providers.  These would access, accumulate, process, etc. the medical records on behalf of patients.  This makes the access problem a whole lot easier, because the access standard does not have to support authenticating individuals directly but rather only the accredited providers.  The accreditation standard should be high enough to avoid fly by night operators that would rip off patient data, but low enough to allow startups to compete.  One could think of these organizations as serving a role similar to the one registrars have for domain names.

I very much hope that the investment from the stimulus is made in establishing these two: access standard + patient record provider licensing.  This would crack the healthcare system wide open for innovation!