Seth Levine's VC Adventure

Revenge of the database

I had a note from a break-out session I led at defrag a few months ago that read "database is back".  It was by far the biggest take away from the two-day conference for me.  While a significant infrastructure has developed around simplifying and virtualizing pretty much every aspect of the technology stack, the common denominator to all NextGenWeb, Web 2.0, social networking, aspiring platform companies is the database.  And while the other elements of the technology stack are getting all of the fanfare the very unsexy database that back-ends all of this great new stuff is the real hero.  After all, many of the companies in the categories I mention above are really just fancy front-ends to a large.  This presents problems for companies that are developing new services since there are very few options for lightweight databases and essentially no options for virtualizing these databases (at least nothing very robust and scalable).  For the most part they're stuck handling the set-up, implementation and maintenance of this technology themselves.  The result is greater cost, more headaches and an inability to quickly scale if their business is successful.

The post I intended to write after the conference was going to point this out and push for a forward thinking company to come up with a solution.  Procrastinate for a few weeks and that's just what happens (minus the original blog post).  Amazon today is releasing a limited beta of SimpleDB – a cloud based database to compliment its S2 and EC2 offerings.  From the Amazon site: "This service works in close conjunction with Amazon Simple Storage Service (Amazon S3) and Amazon Elastic Compute Cloud (Amazon EC2), collectively providing the ability to store, process and query data sets in the cloud."  The addition of SimpleDB makes the Amazon stack an unbelievably compelling value.

This is great news for web companies looking to buy all of their infrastructure by the drink and move infrastructure costs from upfront CapEx to more manageable operational expense.  Not to mention the added ability to quickly scale up or down depending on success.  Sure – eventually when they're successful many companies will find it cheaper (and operational expedient) to manage their own infrastructure.  Eventually . . .

See also:

TechCrunch article on the announcement

GigaOm's take

The missing social network

Facebook trying to co-opt the web into Facebook highlights for me how backwards the social networking world is today. I'm a fan of the platform idea, but the more I think about this, the more I come to the conclusion that the world already has the greatest platform yet developed at its fingertips – the Web itself. I understand why Facebook is trying to enable the reporting of external content all over their site but what would really be great is if rather than trying to port the net into my social network, my social network extended onto the net. When I'm in Facebook, I don't really care that much if Brad just bought Book A or if Chris just purchased movie tickets. When I really care is when I'm on Amazon looking for my next read or when I'm at Fandango about to buy tickets to see a film. More generally, when I hit a site, I'd love to know who else I know who has been there, where they surfed to afterwards and what other related sites I should be checking out based on the behavior of my network. When I'm on a blog, I'd love to see who I know who has left comments, how they've rated the content and if I'm reading something written by someone that's a few degrees of separation from someone I know. I'd like all of my networks (my contacts, my LinkedIn connections, my Facebook friends, etc.) to be a part of this extension of the web and for it to inform and enhance my surfing.

Adaptive Blue is starting to cover some of this with their Blue Organizer. So is me.dium with social browsing and SocialThing with its network of networks. Now we just need to bring it all together. . .

Widgets are s-l-o-w-i-n-g m--e d---o---w---n

The great thing about having a bunch of widgets on my blog is that every time my site slows down I have my choice of people to blame. Shame on me for having so many widgets, I guess, but really - there has to be a better way of managing this.

From start to finish, the experience of inserting a widget on my blog is unsatisfying. I have to configure each separately (so it's hard to make them consistent in look and feel); I have to manually insert the javascript on my blog; to change attributes or location of the widget I have to mess around with the code again or have to go back to the site where I created the widget in the first place; I can't create a library of widgets and turn them on and off at will; and my widgets are constantly breaking something on my site (apologies for those using my Lijit search last week, which I managed to inadvertently disable when I was moving a few things around on my sidebars). And because I use advanced templates in Typepad everything that's designed to make at least some of this easier doesn't work.

On top of all of this, since pages load sequentially, if I have a slow widget or two everything comes to a crawl (and when I say "if", I mean "when . . . daily"). This isn't how the world should work. How about someone developing a CDN for widgets? I'm imaging a widget distribution system that manages the calls back to the widget creators, cashes the appropriate material, allows for better compatibility across browsers and platforms and manages the page loading in a more intelligent fashion (ideally all of my content should load first, then the internal links such as my recent posts, and lastly, all the other stuff I've littered the page with – although I understand that what I'm describing won't be able to do this entirely). This same system could also handle the creation of widgets, streamline how I turn them on or off, how I move them around my site and when they are displayed (i.e., I could rotate them if I wanted to or vary their display based on post topic or blog category). This system would be the single place that my blog called back to in order to render all third party content. If a widget wasn't loading properly, they could just skip it and move on to the next one without holding up the loading process. They'd also be in a better position to architect a network more suitable for this type of content distribution (since most widgets are created to extend content where the widget creator isn't particularly focused on their distribution network once their content is in place).

We've been talking about variants of this in the Boulder tech crowd for months – let's get on it!

BioWest with a ‘Vue

One of the best parts of being a part of TechStars this summer has been watching the companies make their way onward and upward now that TechStars 2007 is fading into memory. While most are still in some form of beta (meaning that their sites don't tell you much about what they are up to) a good handful are making real progress. A great example is EventVue (also still in beta), which provides conference attendees a way to interact before, during and after a conference (which is really the main reason anyone attends a conference in the first place). BioWest just announced that they would be using the platform to support their upcoming conference and my friend Adam Rubenstein (who writes a blog about the Colorado life science investing) has a nice post up on why BioWest went with EventVue. Go TechStars Go!

The gentrification of FaceBook

I first signed up for FaceBook about 8 months ago, interested in checking out the platform and to playing with the technology. I was happy to find a handful of my geek Boulder friends already there as well as a few colleagues and one or two college buddies. The TechStars guys (most of whom were under 25) were, of course, well established on FB already and happy to point me in the right direction (download this, pull your blog into your news feed, see your contacts this way, etc.).

Over the first few months I noticed some of my VC friends join up as well as more of the Boulder tech crowd. Then I started seeing a few more people from College and finally even some friends from high school 'friending' me. I've noticed over the last month or so that this trend is accelerating . . . quickly. And it's not just older tech geeks – it's anyone and everyone starting to jump on and sign up. While the overwhelming majority of FaceBook users are still college and just post college age (18-24) there's real growth happening in the 34 and up crowd (FaceBook itself reports that its fastest growing segment is those over 25, but doesn't break it out more specifically).

What does this all mean? I believe that at the moment, FaceBook is the most important platform on the internet. And while I'm sure that iGoogle and others will give it a run for its money, anyone looking for an audience online can't ignore the 55 million and growing number of users on FaceBook (still doubling every 6 months). And while there are over 7,000 FaceBook applications, the growing number of older . . . I mean experienced . . . users suggests some different ways that publishers and app developers can and should target this demographic. This has implications for gaming, for recommendation engines, for book and movie advertising and reviews, for conferences, for the guys spamming my inbox incessantly with advertisements for Viagra and for just about anyone else who has a product or service to sell for the 35+ crowd.

The beauty of FaceBook as a platform is how broadly applicable it is to any demographic – the utility is a function of the people you are connected with rather than the application itself. This explains why we'll see FaceBook usage accelerate, particularly across new demographics.

They get it

There's no question that OpenSocial supports exactly what I was talking about last week. Check out Marc Andreessen's post on the effort here – a must read for anyone who cares about open platforms.

Web2.0 social-networking SaaS is the way to go!

Someone joked with me the other day that after a recent experience trying to get funding for an old school enterprise software business they were going to reposition themselves as a Web 2.0 social networking SaaS company to see if that helped. Ahh . . . bubble humor . . . it would be even funnier if it didn't ring so true . . .

AppExchange is the new black . . . again

The big news coming out of Web 2.0 (other than Facebook's Evan Zuckerberg somewhat obliquely announcing that FB is going to create its own ad network and dodging a few questions about their developer agreement – see coverage here) was MySpace following in the steps of Facebook and opening up their site to developers. It's no secret that MySpace has struggled – especially relative to Facebook (at Web 2.0 Murdoch also lowered guidance for MySpace revenue and there have been many reports of MySpace's significantly lower growth rates than that of Facebook). Facebook's decision this May to create a standardized development platform for 3^rd party app builders has without question been a huge success. MySpace has now jumped on the platform bandwagon and will begin to better support their developer ecosystem.

I wrote a post about this 18 months ago (see AppExchange is the new black from June 2006) applauding companies such as eBay and SalesForce for cultivating their developer networks. While Metcalfe's law suggests that the power of a network increases exponentially with the number of nodes on that network, there is perhaps an adjunct to Metcalfe's which would state that this power increases by another factor as you add functionality across that network. Think of the programs developed for a networks as additional nodes, but where the number of 'nodes' to factor into Metcalfe's equation equals not one for each new program, but one for each program multiplied by the number of users it has. Networks that serve as platforms become ecosystems. Networks that don't are silos.

So what's next? In my June 2006 post about I referenced Oracle (and didn't but could have referenced SAP) – they are both migrating down the stack and building systems that are both applications and middleware. They should start thinking about how to better create an ecosystem around their systems rather than continuing to believe they can develop (or buy) all the functionality their customers need. Thinking a little bit more broadly, the Yahoo, Microsoft, AOL and Google all have large numbers of email users that are huge potential networks and which are completely untapped (see this WSJ article from last week; hat tip to Ric for the pointer). Without question the social networks are adding messaging and mail like functionality to their systems – the mail guys, with a wealth of data about their users and with a more broadly 'sticky' application, should start looking to fight back. Perhaps we'll see the first of this with iGoogle.

I'm definitely a platform guy – I see the power of building a foundation and having others add to it. And while I may be overstating the future opportunities here, the predictions I made 18 months ago and the moves by the large networks since then would suggest otherwise.

TechStars Rocks!

Wow! What can I saw about the TechStars investor pitch/demo day except that it was fantastic. The teams were extremely well prepared and every single one of them really nailed it. We had about 50 private investors and venture capitalists in attendance to see the 10 TechStars teams go through what they've been up to this summer.

Those of you who follow me on Twitter know that I've been somewhat of a TechStars groupie this summer – I've spent many hours with the teams giving feedback, advice and lending a sympathetic ear. I've seen early ideas and kluge demos turn into the seeds of real business ideas, real prototypes and true products. The journey has been amazing to watch. For me this was the best part of yesterday – the progress that each of these companies has made this summer is truly remarkable.

TechStars was raw entrepreneurship at its finest. The program brought in teams who had ideas that ranged from pet projects to recently crafted plans to companies that the founders had been working on already for years. Over the course of the summer these ideas were refined, changed, thrown out completely and refined again. The result was a pretty amazing experience for the founders and ten pretty interesting potential businesses.

Don Doge had a nice summary of each on his blog if you're interested. I've also linked to the companies below.

EventVue

IntenseDebate

SocialThing

StickyNotes

Search-to-Phone

Villij

MadKast

FiltrBox

KBLabs

BrightKite

Monday morning

As promised – a few thoughts on StartupWeekend now that I've had a chance to both get over my disappointment that we didn't release anything on Monday morning and more importantly time to think about what worked and what didn't work last weekend. There seems to be no lack of opinion on the subject (see the comments to the "Success and Failure" post on the StartupWeekend website and the mass of comments left on the TechCrunch article about the weekend or just do a Google search for StartupWeekend).

Overall, it was a great experience. We had an outstanding group of founders from a broad cross-section of Denver/Boulder technology companies. Andrew did a great job of facilitating the weekend without being too heavy handed (although see below for some ideas on where more structure might have benefitted the ultimate goal of the project). A lot of the people involved have posted on their blogs or elsewhere online that as a "Social Experiment" StartupWeekend was a success. I think what they mean is that we had a fun time, everyone contributed, and no-one wanted to kill anyone at the end of the weekend (which says something given that many people were operating on only a few hours sleep over the 50+ hours that the project was active). That said, if StartupWeekend is only a social experiment the concept will have pretty limited value (and a limited life – it's fun to do once, but to make it a regular occurrence, there needs to be the realistic expectation that a 'product' will actually be released). Don't get me wrong – I loved the 'social experiment' side of the weekend, but the goal of the project was to produce something in the period of time we were together (and capitalize on the buzz generated by the excitement around the process for the benefit of the company that was created). The journey was great, but we actually needed to get somewhere. I know Andrew wants to replicate the Boulder StartupWeekend elsewhere around the country - I'd love to see this happen and think what we learned last weekend will help make the effort in other cities more productive and ultimately more successful.

That said, VoSnap will launch – and soon. The project did live beyond the weekend and, after everyone got some sleep and returned to their Monday-Friday lives, there was a great deal of interest by the group in continuing with the project. Not just to the first launch, but beyond that (and potentially building a business around the project). When it's up and working I'll put up a note and link to the working site. It definitely says something both about the weekend and the group of people who participated that the project will live on beyond Monday. It's emblematic of why the weekend was such a great experience.

Here are some summary thoughts on what worked, what didn't work and what we might do differently when we do this again:

What worked:

• Picking the idea: We ended up with something that was realistic to do in a weekend and had the potential for legs beyond the two day project. Having a site where the founders could post ideas and doing an initial vote before we showed up on Friday night were keys to vetting the list before we even showed up (we spent around an hour picking the idea – that felt about right)
• Group dynamics: While there were plenty of type-a personalities in attendance, the group self organized really well and it was clear that people checked their egos at the door. It helped that this was a one weekend project (at least that was the thought going into it). A lot of work was done by these groups – much of it not visible to people following along on-line. PR, business development, marketing, etc. were all on their game and came up with plans/ideas that will serve the company well beyond Sunday night.
• Focusing on one project: When I looked around the room on Friday night I thought to myself "no way can this many people stay busy for the weekend". Andrew held to his guns and insisted that we all work the same idea. He was right on – while a group of much more than this size probably wouldn't work the weekend needs to be about ONE idea and the focus of the group needs to be singular in that respect.
• Quick meetings/quick decisions: Every hour we met for 7 minutes for an update. This is an incredibly effective way to communicate with a group w/o disrupting everyone for a long period of time. A few people have actually extended this idea beyond StartupWeekend to their day jobs. We were maniacal about keeping these short and as a result there was a lot of information conveyed at regular intervals in a very short period of time. Andrew also implemented a system of "quick votes" to make decisions (VoSnap would be perfect for this!) – the idea was to lay out the options, vote and move on. Quick votes allowed us to keep moving throughout the weekend.
• Buzz. StartupWeekend really took on a life of its own online last weekend. It started with David's regular blog posts on www.startupweekend.com but also included the live video stream that was put up (which regularly had nearly 100 viewers), flickr photos and, of course, the TechCrunch posting. At the beginning of the weekend very few people outside of the Boulder tech community knew what StartupWeekend was and VoSnap didn't even exist – by the end of the weekend there were hundreds of mentions online of both and tens of thousands of hits to the VoSnap site (there are more than a thousand people now on the VoSnap mailing list).

What didn't work/what to do differently:

• The product. I guess I have to start here. On Monday we had a lot of things working, but the product wasn't one of them (still isn't – at least not publically). I think we should have put up whatever we had on Monday morning. On the one hand, this would have been fodder for those online who were down on the idea of creating a product over the weekend. On the other hand, it would have been true to the idea of the weekend (to create something in 2 days).
• The development process. We had a lot of great developers, but we made a bad choice of dev environments (at least in the context of putting something out in a weekend) and as a result we ended up scrapping a lot of the initial work that was done on Saturday on the product. In retrospect, we should have appointed a dev lead ahead of time, probably had the development team meet before the weekend to review the likely top candidates (based on the pre-weekend voting) and make the choice of development platforms ahead of time. The decision wouldn't have been rushed and it would have been more well thought out. In the end, we made some choices that might have been right for the long term success of the product, but made it difficult for us to complete the first rev in time to launch on Monday. There were also a few among us that recognized this early (David, Andrew and I had this conversation on Saturday) but we (wrongly) decided to let it go.
• More developers. Out of 70 people we had probably 7 hard core developers. This ratio was too light. Many of our early stage companies are between 30% and 50% development staff. While on top of the 7 dev types, we had a handful of other people that would probably count as "engineering" in the numbers I just quoted, this imbalance set us up for some challenges (not the least of which was that development of the business plan, pr, marketing, etc pretty quickly got very far ahead of what we could do in a short period of time and probably contributed at least some to feature creep).
• Communication after the fact. During the weekend we did a great job of communicating with the outside world. Through emails, blogs, streaming video, flickr, etc, we were very open about what we were up to. A surprising amount of work has taken place since Monday, but we've been totally mum about it beyond the founder group (to give you a sense of the volume of email traffic, I've created a separate inbox just for VoSnap related email so I can see it all in one place).

I'll end with a few pictures to give you a sense of the environment we were working in (two views of the main room and one of a stand-up product meeting).