Continuations

I have met Senator Patrick Leahy in person several times and owe him a personal debt of gratitude.  Over the years I have been impressed by how he has conducted himself in politics.  He is, however, dead wrong in his introduction of the Combating Online Infringement and Counterfeits Act (COICA) act together with Senator Orin Hatch of Utah and sadly also co-sponsored by New York’s Chuck Schumer.   COICA essentially establishes Internet Blacklists and gives ISPs immunity when they enforce the blacklists, meaning you can’t sue your ISP for suppressing a site that is on the blacklist.  The current version of the bill proposes two separate blacklists: one that can be added to only by courts and another that can be added to by Attorney Generals.  Sites would be added, according to the bill, if they are dedicated to infringing activity, such as making lots of copyrighted material available for download. 

Everybody should read this great summary of the bill and why it is a bad idea.   After reading that I would encourage you to sign the petition to stop the bill (on the same page).  I also encourage everyone to check out this powerful view of who has been lobbying in support of COICA.  If you have been paying attention to this set of issues, the name of the bill will also ring a bell.  With its use of the word “counterfeits” it is clearly linked to the equally misguided international effort known as the Anti-Counterfeiting Trade Agreement (ACTA), which I have decried previously.

Now just to give you some idea of how incredibly bad an idea this is.  Imagine a future conversation between Attorney Generals and oh say Craigslist.  AGs:  “We’ll have you added to the Blacklist”  CL: “We are not infringing any copyrights” AGs: “You can fight that in court - *after* the entire country has lost access to your site.”  The power of threatened law suits alone was enough to get Craigslist to shutter its perfectly legal adult services section — now imagine the change in the balance of power if there is a way to eradicate a site from the Internet.

If you care about freedom and democracy you do not want to give the government a wholesale way to shut down access to sites on the Internet.  The potential downside from abuses of such as system far outweigh the upside to copyright holders.  We badly need an Internet Bill of Rights that codifies basic notions of freedom of access so that we don’t have to fight this fight over and over again.

Unlike many other people, I have not been disappointed with Barack Obama’s presidency.  There have been many moments of frustration, but it is a damn tough job, especially when so many representatives from both sides seem to have lost all sense of what their responsibilities are.  But I do find the idea of assassinations targeting US citizens without due process deeply disturbing.  Due process is one of the key breakthroughs of the modern state and is at the foundation of what sets us apart from dictatorships.  Now I do believe that it is possible to have situations where US citizens pose such a threat that deadly force is warranted.  Clearly that is already the case today — for instance, when the police confronts an armed robber.  I can even see that this logic could be extended to US citizens living abroad and actively planning a terrorist strike.  But not in an executive branch only fashion.  That is a step back to the Middle Ages.   Here is an appropriately outraged piece with lots of good links at Salon on this subject.

Excited to be announcing our investment in Shapeways and the relocation of the company’s headquarters to NY.  You can read more about it on the Shapeways blog and over at the NY Times.  The founders Peter, Marleen and Robert are relocating from the Netherlands — welcome to NY!

Mike Arrington kicked off a tempest in a teapot yesterday with his harmlessly titled post “A Blogger Walks Into a Bar” that alleges a price fixing conspiracy for angel investing.  I don’t claim to know what happened at the meeting, but I fully concur with Fred’s assessment that there is no evidence of any fixing (only the opposite).  Also, the idea that a small number of individuals could collude meaningfully in the early stage market — even if they wanted to — is bogus to begin with.  There are way too many participants in that market for that to ever work!

Finally, I do know at least two people who participated well - Bryce Roberts and Dave McClure.  We are co-investors with both and they are terrific champions for entrepreneurs.  Which, by the way, doesn’t mean that they won’t sometimes complain about valuations - we all do!  You can and should read Bryce’s comment on the matter here and Dave has a great post.

We are excited to be investors in Twilio as they transform telephony from cumbersome, proprietary and expensive into an easy-to-use web service.  We also love the team’s creativity in promoting Twilio to developers, including the idea for this week’s contest: lunch with the USV team.  We will be working with the Twilio team to pick the winner from this week’s entrants.  As such, I thought it would be helpful to point to some of the criteria that we look for in evaluating innovation on the web.  Fortunately, this is an easy task for me as we have written about it extensively on the USV blog.  Since much of these points were written some time ago, I was pleased to see that we still believe in them as much if not more so than when originally posted. 

Two points that are conspicuously absent from our list are the technology itself and features.  And that’s with good reason.  We believe that on the web technology provides little or no sustainable competitive advantage.  First, when something is delivered over the web, endusers tend not to care about how it’s done (which is different from installed software, which had to fit into someone’s existing environment).  Second, your competitors can generally see what you are doing in terms of features.  So if you are competing on features, they can add similar features quickly.  So as we look at projects for this contest we will not care about whether you wrote it in Java, PHP, Python, Perl, Ruby or something else altogether (Scala? Clojure?).  We will also not look for whether you have more features than another project.  

Instead, we will look at how you have leveraged Twilio to create something that could have a network effect or result in accumulating a data asset.  We are looking forward to see what folks come up with!

I continue to be amazed at the things we have available to us - we generally live in the future.  But every once in a while the past intrudes with full force.  As in when I tried to connect an HTC Evo to a Mac laptop with a USB cable to transfer a bunch of video.  At first, I got absolutely nothing.  Then I went into settings on the Evo to discover that there is a setting to change the USB connection from “charging only” to “mount.”  Easy, I thought, until I noticed a pop-up on the Mac saying “Unrecognized volume” with three options “Format”, “Ignore” and “Eject.”  I very gingerly moved the pointer so as to not accidentally hit the “Format” button and gently tapped “Eject”, sighing a (pre-mature) sigh of relief. 

Turning back to the Evo, I suddenly see an unhappy SD card icon flashing.  I go to camera to find that there is no archive — not a trace of the video.  At this point, I am in a minor state of panic at the thought of having lost the video that Susan and our youngest spent the morning recording.  Again, this time on the Evo there is the helpful suggestion of letting me reformat the SD card.  Not quite willing to give up, I turn the Evo off, take out the battery and reboot.  Going back to the camera, the Evo initially still shows no sign of the videos, but then flashes as “Searching for images” on the screen and after doing that twice finally recovers the videos.

At this point I search around and find that other people have run into this problem.  Someone provides a solution using USB debug mode but I couldn’t get that to work either.  In the end, I wound up transferring the videos via Bluetooth which is quite slow compared to a USB connection.  It seems incongruous that one could have a device as powerful as the Evo and then have it stumped by a seemingly trivial problem.  This whole experience reminds me of something else that never seizes to amaze me: engineers can make airplanes fly but they can’t get the cabin temperature to stay in a comfortable range.  Oh well.  Maybe it’s good to have some reminders of the past.

As I wrote last week, I created Preditter as a project to use some of the technologies that we have invested in.  I blogged on Monday about how amazingly easy it is to use the Twilio API.  Up today: MongoDB (which has been created by the team at 10gen).  Here too, I was blown away by the ease of getting going.  You can literally be up and running with MongoDB in minutes (on my Mac laptop I just downloaded and on my Slice running CentOS I used yum).   But the thing that I was most pleased about is how truly effective the document model is in supporting agile development.

I remember how on one of the early releases, I wrote to Kristina about the PHP driver, asking why it was based on using arrays as opposed to PHP objects.  Kristina rightly pointed out at the time that arrays are far more flexible and that others can construct object mappers on top of arrays (which is exactly what has happened with several active projects).  But as it turns out — at least for me — developing with arrays, instead of full-blown objects is also a great way to get started on a project. 

I wanted to make most of the code for Preditter be generic, as opposed to NFL specific, so that if I wanted to it would be easy to add other categories later (e.g., the upcoming elections).   In Preditter, predictions are about events which are stored in an events collection.  The beauty of the document model and array based development is that the only fields that all my events need to have in common are things like category (for now: sports) and subcategory (for now: football) plus a tag array (for such things as “regular season”, “nfl”, etc).  Everything else can vary across events with fields that make most sense for the specific category and subcategory.  Instead of having to pre-plan for an object hierarchy and think through members and inheritance and methods and all that, I can just sling around arrays through most of the code and dispatch to specific handlers as needed (based on category and subcategory).

But it gets better.  In the predictions collection instead of referring to the event by a foreign key (as you would in traditional relational design under normalization), MongoDB makes denormalization a cinch.  Every prediction simply includes as an embedded document the event that it refers to.    The code for doing that can be entirely generic!  It just stuffs an event into a prediction and doesn’t need to know a thing about what’s inside the event.   Similarly, when retrieving predictions — for instance all the predictions made by a specific user — the code for doing so can retrieve the associated (read: embedded) events without any knowledge about the type of event that the prediction is about.

But wait, it gets even better.  The embedded documents are first class citizens when it comes to querying!  For instance, finding all the predictions that are about a game involving a specific team or on a given date (or what have you) becomes a single query against the predictions collection that looks inside the embedded events (and you can have secondary indices to make those queries performant).  So here the true power is revealed.  Not only does MongoDB superbly support agile development, but it simultaneously allows for scaling.  Should Preditter unexpectedly take off wildly, I can even shard the predicitions collection without touching a single line of code!

Last week, I participated in a panel on the implications of SaaS/Cloud for SMEs.  One point that was made by several of the panelists is that a big advantage over on-premise software is that you are always on the latest version and thus get to benefit from improvements over time.  Those improvements can be additional features but they can also come in the for of reduced prices.  Amazon for example has cut their prices for EC2 and also made such cost savings available as reserved instances and most recently micro instances.   As Amazon’s resource pool grows they can drive down the cost and they are passing some of those savings on to their customers. 

Our portfolio company Twilio is taking a page from the same playbook.  They just announced new pricing that will result in cost reductions of up to two thirds for some applications!  In Twilio’s case too, the advantage of customers from participating in a cloud solution is that  everyone’s resource demands can be pooled allowing Twilio to drive down cost overall and pass some of these savings on.  Expect more of the same as Twilio continues to grow!

For all developers and entrepreneurs: Twilio is offering a creative contest prize - lunch with the USV team.  So fire up your editors …

So the results are in on the first week of NFL predictions on Preditter:  10 correct calls, 5 incorrect and 1 inconclusive (tweets were 50:50).   Not much to write home about, but kind of fun.  Especially since Preditter now shows who was right and who was wrong.  For instance, for one that Preditter got wrong see Jets vs Ravens.  For a correct one check out Saints vs Vikings.  There are also the beginnings of user pages.  For instance, my page is at http://preditter.com/users/albertwenger (these don’t yet show which predictions were correct — coming this weekend together with leaderboards). 

Games for Week 2 are loaded, so preditter away!

I have previously written about non-linearity and how it messes with our reasoning.  In business non-linearity abounds.  One example that has been discussed a lot is the change between going from a price of 2 cents to 1 cent compared to the change with going from 1 cent to free.  Twitter wouldn’t be twitter if it had charged just 1 cent per tweet.

One area where this is also true but more subtle and less well understood is product simplicity.  Steve Jobs is clearly on to something very powerful by insisting that the iPhone (and now iPad) have exactly one button on the front.   The iPhone’s simplicity went beyond that initially by launching with a bare bones version of what could be done on the phone (remember: no copy-paste).  The simplicity was at the heart of a huge non-linearity — the iPhone rapidly got many users which attracted developers setting off a self-reinforcing cycle.

The same can be true for something as technical as APIs.  The simplicity of Twilio’s API is such that adding previously complicated telephony applications becomes possible in a matter of hours.  I already knew that from the due diligence we had done prior to our investment, but when I finally got around to using it in an application (see Preditter or text “Jets” to (585) 466-4919 before the game starts tonight) I was still completely blown away.  From the point of hitting the “Get Started” button on the Twilio home page to being to having the functionality live for the world was less than 10 minutes.

The iPhone was not just a simpler Windows Mobile phone.  It was so much simpler it was totally different.  Twilio is not just a simpler IVR platform, it is something entirely different.  In each case, the simplicity of the offering unlocks entirely new set of users and use cases, which results in a hugely non-linear change in adoption.

It doesn’t really matter whether it is Android’s growth or the FTC’s investigation, but Apple has officially rescinded some of the most onerous restrictions on developers for iOS by allowing third party tools.  This is an important change from a developer perspective as it makes cross platform development easier.  Not just across mobile devices, but also across the web and mobile (of course HTML5 pushes in that direction as well).

Just as interesting though is the increased transparency that Apple is now providing on its store review guidelines.  Engadget has a terrific summary with a link to the actual PDF.   Apple’s model for the app store still has many elements of a more traditional publishing model.  Apple clearly feels that they need to curate to some degree what’s in the store (“no more fart apps”).  But why?  What is wrong with letting someone create another fart app?  Absolutely nothing, as far as I can tell.

I believe that ultimately an open publishing system for apps will win out over a closed one.  That system, however, needs to crack the nut on making it easy to find the best app(s) for a given purpose without running across a lot of garbage along the way.  Apple is a far way from this.  The other day I was looking for a calculator for the iPad for my kids to use and had to resort to google as the App Store experience was entirely broken.

When investing in consumer services, we try to be active users of the services.  That tends to provide a fair bit of insight about what is working and what is not (as long as one keeps in mind that one may not be the exact target user!).  To do something similar for our more developer directed investments, I have long been meaning to build a little project using MongoDB and Twilio.  But usually the extremely rare cycles that I have available for code go into fixing something on DailyLit.  Over Labor Day weekend, however, I had some unexpected extra time and used that to hack together a little experiment which you can check out at http://preditter.com (warning: very raw at the moment).  It was a terrific learning experience and I will post some follow ups with lessons learned.

This morning at 8am I will participate in a discussion around the costs and benefits for small to mid-size enterprises (SMEs) from using Software-as-a-Service (SaaS).  Here are some of my key thoughts:

1. Total cost of ownership (TCO) studies are highly suspect.  The real cost of installed software has historically been an ossification of organizational structure around fairly inflexible offerings.  With SaaS we are seeing more flexibility around what can be adopted by whom and how it can change over time.  It is virtually impossible to put a meaningful number on that gain in flexibility (it represents some kind of “option value,” which is notoriously hard to get at).
   
   
2. We are still in the relatively early days of what SaaS will ultimately bring.  The best analogy I have is that a lot of SaaS still feels the equivalent of Web 1.0 on the consumer web: the existence of other users of the same software is hardly noticeable rather than being a pervasive aspect (never mind across companies, even just within the same company).
   
   
3. The business models enabled for software providers by SaaS are similarly in their infancy.  Most are still charging for the software, albeit now on a usage basis.  But Freemium is beginning to make significant inroads and the final frontier is free SaaS (such as the PracticeFusion ERM) with money being made off transactions that take place through the system.
   
   
4. Offerings from different vendors are still too difficult to integrate.  Maybe this is endemic to software because of incentives by vendors to try to remain closed, but I think the logic of web services suggests that leaner more focused offerings with more scale will ultimately succeed.  SMEs won’t necessarily have the skills in-house to do the integration, which provides an opportunity not unlike the original vision for GrandCentral (which was a decade or possibly two too early and had to pivot radically before becoming GoogleVoice).

If there are any great insights that come out of the discussion, I will either add here or write another post in the near future.  For now I have a train to catch to the city!

Craigslist has self-censored its Adult Services section in the US (instead of its usual location the word “Censored” appears and the section is no longer accessible).  There has been a lot of discussion of this move already.  Two pieces in particular are worth reading.  Jeff Jarvis points out that one of the reasons why Craigslist is under attack (and possibly the main reason) is just how disruptive it is to existing media businesses.  Danah Boyd - guest writing at the Huffington Post - provides a pragmatic defense based on the benefits of transparency and a centralized location for law enforcement.

I have long admired Craigslist as one of the first truly net native businesses.  Whether by accident or by design, Craigslist has chosen to maximize the size of its marketplace and charge only a fraction of its users for a fraction of their activities.  That has made it a formidable competitor in the classified advertising space.  Their success over existing media businesses was made possible by net neutrality — the fact that access to Craigslist’s servers is as fast and unencumbered as that to those of much bigger companies: Craigslist has fewer than 50 employees.

Therein also lies an important problem.  Craigslist does not have the resources to review every listing.  The original section in the US was titled Erotic Services (as it still is in other parts of the world, e.g. the UK).  Under pressure from several AGs, Craigslist replaced this with Adult Services and started charging a listing fee to generate the revenues to cover the expenses of examining listings in that category.  Yet that still was not good enough to keep the AGs at bay.

So the upshot is that as the fight to protect existing legal protections (the DMCA safe harbor) and de facto regimes (net neutrality) heats up, it is important to watch out not only for legislative initiatives but also for misguided enforcement.  Because otherwise it will be Craigslist today, Blogger tomorrow.