Following the news that Yahoo will acquire social sharing platform Tumblr for a massive $1.1 billion in cash, it’s becoming clear that two of these investing svengalis are Bijan Sabet of Spark Capital, and Fred Wilson of Union Square Ventures. They’re the venture capital rockstars of the social web as their jaw-dropping track records shows — they have been very early investors in companies like Twitter, Tumblr and Foursquare. Wilson was also early to Zynga, Etsy and Kickstarter.

David Karp, Founder and CEO, Tumblr paidContent Live 2013 Albert Chau / itsmebert.com

We’ve seen this pattern before. During the networking boom of the late 1990s, it was Vinod Khosla and Promod Haque of Norwest Venture Partners who scored many of the top deals. Khosla was early to the table with companies like Cerent, Siara and Juniper Networks.

For the first wave of the consumer Internet it was John Doerr who owned it all, and later during the search boom, he and Sequoia Capital’s Mike Moritz led the charge by investing in Google. Similarly, the shift to the social web (or post Web 2.0) has been a happy hunting ground for Wilson and Sabet.

While there are others such as David Sze of Greylock and Jim Breyer of Accel Partners who have found success in social with Facebook, the Sabet and Wilson team has a better on-base percentage. And one of the reasons why these two guys have found gold faster than others is because they are active participants in the social web and invest in what they know.

Sabet, who moved up to number 51 on Forbes’ Midas List in 2013, was described by Tumblr’s founder and CEO David Karp as someone who “lives and breathes,” social media, and as Karp’s “mentor, friend and partner from day zero.” Sabet said in a post this morning that he met Karp when he was 19 and “was immediately taken with his passion and drive to create wonderful things.” His other investments include Twitter, Foursquare, OMGPOP (sold to Zynga for $180 million), and thePlatform (acquired by Comcast), as well as RunKeeper, Boxee, Jelly, Lyft and Stack Exchange.

Wilson, who writes the popular blog AVC, is one of the most winning venture investors thus far in the new century. He rose to number 16 on Forbes’ Midas List in 2013. In addition to Tumblr, Wilson has backed Twitter, Zynga, Etsy, Kickstarter, FeedBurner, Lending Club, Zemanta, ComScore and Tacoda (sold to AOL in 2007).

In a sign of the new power of the New York startup ecosystem, Tumblr is run out of New York, and both Wilson and Sabet are East Coasters — Wilson in New York, Sabet in Boston — but of course spend a lot of time in Silicon Valley. It’s through these two investors, who live across the country from Sand Hill Road, that a major chunk of the web’s leading social brands have emerged.

Pulling out lessons from these master’s investing practices, would be a good topic for a book. But I’ll run through just a couple I see that are obvious. Sabet clearly has an eye for finding the passionate creator founder early, and has an ability to see the future. Wilson has long practiced what he has preached, is utterly immersed in social and has a unique way with words. He writes this morning on his blog about one of the cliches of VC investing: “success has a thousand fathers.”

Clearly Sabet and Wilson are two of the main fathers of this morning’s success story.

Updated at 1:20 PM on May 20, with comment from Fred Wilson. Wilson attributed some of the deals to his Union Square venture partners and says that “investment syndicates are a team.”

Related research and analysis from GigaOM Pro:
Subscriber content. Sign up for a free trial.

Related research and analysis from GigaOM Pro:
Subscriber content. Sign up for a free trial.

• Social media in Q1: commerce and discovery dominated
• NewNet Q4: Platform mania and social commerce shakeout
• Big Data, ARM and Legal Troubles Transformed Infrastructure in Q4

Related research and analysis from GigaOM Pro:
Subscriber content. Sign up for a free trial.

• Who and what to watch in the new era of the living room
• 2012: Data, spectrum and the race to LTE
• CES 2012: a recap and analysis

As great as shared cloud infrastructure can be, these workloads often demand dedicated physical servers in addition to heavily virtualized, shared cloud servers. That’s where SoftLayer, the Dallas-based cloud service provider, says it differentiates itself from other big cloud players including market leader Amazon Web Services. SoftLayer, as Derrick Harris reported earlier, may not be a household name, but it runs the cloud that helps run many household brands including Tumblr, Cloudant and Citrix and has backed such popular games as OMGPOP and Draw Something, which have since been acquired by Zynga, which is moving more of its internal games onto its Zcloud and off of AWS.

SoftLayer says: hybrid cloud is best for gamers

SoftLayer attacks this online gaming challenge with its own hybrid technology which pairs shared and dedicated infrastructure.

“With gaming either on mobile devices or on Facebook, it’s all about engagement. How much time can you keep players engaged and coming back and what drives all that is the data points collected by the games makers that help them refine and iterate fast,” Marc Jones, VP of product innovation for Dallas-based SoftLayer told me last week. “One driver from an infrastructure perspective is the sheer volume of data points [from user moves] which ultimately end up in a database, maybe in MySQL or one of many NoSQL options like MongoDB or Hadoop,” Jones said.

That proliferation of data points and the need for fast response drives demand for higher I/O that Jones says is only available from truly dedicated servers. “With those servers you can configure your drives, the CPU you want, the memory you want up to 10G — you can use local SSD RAID configurations if you want to get better I/OPS and better I/O,” Jones said.

Meeting the fluctuating demand of gamers

The spiky nature of demand for games is another factor that makes SoftLayer’s hybrid approach effective, added George Karidis the company’s chief strategy officer. When a game launches there is usually a spike — the size of which is hard to predict.  ”The variability of game play, the time of day and when they launch means they have to leverage their virtual cloud instances and spin up and down fast.  WIth our private network you can have your cloud instances and your dedicated infrastructure all talking to each other on our private network,” Jones said. That eliminates latency.

Current SoftLayer gaming customers include Kixeye, the San Francisco game company behind Backyard Monsters, Battle Pirates and War Commander.

SoftLayer, other clouds, take aim at Amazon

SoftLayer is getting aggressive in its promotion of itself as a viable — and oftentimes less expensive — alternative to the Amazon cloud. “Our private network is completely free to these users. We only charge for outbound traffic — inbound and data center-to-data center traffic is all included. On a dollar per dollar basis we’re less expensive than Amazon when you go above their low-end compute offering,” said Jones.

Indeed, other competitors are also getting feisty in positioning themselves against Amazon. Last week, Steve Tuck, GM for Joyent, another big cloud service provider, took aim at Amazon in a blog post on Computer Business Review saying it is no longer impossible to imagine competitors (like, um Joyent perhaps?) besting Amazon at its own game, especially as the “internet of things” comes online.

According to Tuck’s post:

Once upon a time, it was easier for developers, entrepreneurs and IT managers to work with Amazon. That is no longer true today. While Amazon remains the dominant cloud provider, its customers are beginning to see cracks in the Amazon Cloud foundation.

Clearly, SoftLayer, with its 13 data centers worldwide and 100,000 servers under management, feels the same way. Paul Santinelli, general partner at VC firm North Bridge Venture Partners, who watches this space closely, agreed that Amazon is by no means invulnerable to scrappy competition.

“That space is just heating up. The frustration with Amazon on the customer side is reliability. ‘You tell me I have servers but they may not be reserved for me and they may be in a distant data center so latency is problematic.’ If you, as a customer, want dedicated instances, reliability and lower latency, it’s highly likely you’ll go to a smaller cloud provider, ” he said.

Related research and analysis from GigaOM Pro:
Subscriber content. Sign up for a free trial.

• Migrating media applications to the private cloud: best practices for businesses
• Putting Big Data to Work: Opportunities for Enterprises
• Takeaways from the second quarter in cloud and data

But while Zynga CEO Mark Pincus was all smiles on signing the deal, he may be less impressed six weeks on, thanks to an intriguing report from Forbes that indicates a precipitous decline in the game’s user base since the acquisition.

According to figures from AppData.com, the simple and highly-addictive drawing game — which rocketed to 36 million downloads in just a few weeks — has lost around a third of its daily active users in the past month.

So what’s going wrong? There seem to be three major possibilities on the cards.

Technical problems

Lots of players have reported trouble accessing the game over the past few weeks. That seems to be a scaling problem, with the servers struggling to meet the demand of more and more downloads — though the move to shift OMGPOP’s traffic onto Zynga’s “Z-cloud” may also be a culprit, since similar reports are also surfacing about another game, Words With Friends.

For many new and existing players, server issues that mean dropped games or lost moves are an immediate black mark against the game: something that can really hurt in the instant gratification world of mobile entertainment.

Zyngafication

Moving to the Z-cloud meant something else, too: the addition of new features that seem to have soured many players’ relationships with the game. For the most part, this means pushy Facebook and Twitter integration — though the company does seem to have also expanded the number of advertisements, too, taking more money from Doritos, the NHL and others to include brand-related words into the game’s dictionary.

These might not be a turn-off for everyone, but they do seem to have had an impact on some users who feel uncomfortable at the commercial options.

Natural wastage

While Draw Something was an overnight success — spreading virally, zooming up the charts and bringing in a lot of revenue for its makers — many people are saying that they bored of it rapidly. That’s not entirely untypical for a faddish game, and since it has little variation, a pretty limited dictionary and real no long-term stickyness it appears OMGPOP hasn’t done a great deal to boost the game’s longevity.

This sort of wastage is pretty normal for short attention span games, which makes their success all about growth: you have to keep marketing the game to new players and expanding the overall market to compensate for those who switch off.

So who’s fault is this? And is it a real problem, or just a temporary glitch?

Some may argue that this isn’t a big deal for Zynga, since it owns not just Draw Something, but the entire business and team behind it. However, as we wrote at the time, Zynga’s purchase was not really about talent acquisition — while OMGPOP’s team are no doubt smart, they’d produced so many duds that the company didn’t have many rolls of the dice left.

No, for Zynga it was a numbers game: the revenues and user base of Draw Something were interesting, but underneath it all it was really about gaining access to a large, mainstream, purchase-happy audience.

Those users are great targets for Zynga to sell its other games to. And the speed at which Draw Something was growing — it had more than 36 million downloads in just a few weeks — meant that the audience was going to be significant.

Except it’s not. And that’s got to hurt.

If this drop continues — in fact, even if the decline is arrested — it could become a really serious problem for Zynga. After all, it not only shows a lack of judgment from management in buying into a fad when it’s at its very peak, but it also suggests that the company wants to strip mine new assets rather than build them up.

Because while scaling issues and a lack of longevity may be rooted in problems at OMGPOP’s end, the reality is that all these issues have been made worse, not better, by Zynga’s involvement. It’s either failed to support the game fast enough and bring in enough new users, or it’s been actively making it worse for players.

And none of those are good news.

Related research and analysis from GigaOM Pro:
Subscriber content. Sign up for a free trial.

• Takeaways from mobile’s second quarter
• Social third-quarter 2012: analysis and outlook
• Hyperlocal: opportunities for publishers and developers

During the first quarter of the year VCs invested $5.8 billion in 758 deals, a 19 percent decrease in dollars from the previous quarter and a 15 percent decrease in deals. This isn’t unexpected as the industry has also seen its ability to raise funding from its investors drop after the financial crisis in 2008 and 2009, and because the first quarter is historically the weakest one for VCs. The VC industry moves in five-to-seven-year cycles, so anticipated trends from even a few years back can take a while to be felt, and then they can change on a dime. Or a deal.

But this trend won’t be felt for long, and there’s one huge reason why: Instagram. The sale of that photo-sharing site to Facebook for $1 billion in funny money and cash caused every single venture firm to sit up and take notice. And behind Instagram are several successful exits with LinkedIn, Yelp, Zynga and Splunk going public.

Facebook has filed to raise capital on the public markets and that should lead to more interest in the asset class, plus it will also have the stock to do crazy inflated deals such as Instagram. Zynga’s assertion this week that it planned to make more acquisitions is only the start of an all-out race to the top to be the king of the consumer web. So entrepreneurs can expect more opportunities to raise capital and the anticipated shrinking of the industry won’t last long.

On the funding side, the dollars flowing into seed and Series A deals also dropped to 45 percent of total deal volume, while the average size of such deals increased to $2.7 million. Again, this doesn’t reflect the change in the market post-Instagram and as VCs start doing deals again in the second quarter. In fact, it’s getting frothy out there with some startups commanding crazy post-round valuations of up to $7 million after taking on a Series A/seed round of $1.5 or $2 million.

In January I pointed out that the trends in early stage funding indicated that the Series B rounds were when entrepreneurs were feeling squeezed. VCs were willing to toss money at any deal in the early stages as a way of claiming their territory. However, if those investments didn’t show significant traction, revenue or product development, they weren’t getting their Series B dollars. Venture firms want a Pinterest by the second round, so entrepreneurs should use their Series A rounds prudently.

The hunt for whales at the Series B level concerns Chris Dixon, a co-founder of Hunch who wrote last week:

The problem with this model of Series A and B investing is that, in reality, many of the companies with big hits weren’t overnight successes. Pinterest, OMGPOP, Twitter, and Tumblr were around for years before taking off and all benefited greatly from having patient investors. In the current financing environment, a lot of good companies won’t live to get Series As and Bs and big VCs will pay valuations on hits that are priced to perfection.

The question that remains to be seen is if the Internet has crossed over so totally into a consumer-market story that venture success and returns drop to reflect the crazy competition that’s endemic in the consumer marketplace. A home run is only a home run if your returns are high, and paying too much to get into the round can give a venture firm a publicity win, but less of a monetary one.

Related research and analysis from GigaOM Pro:
Subscriber content. Sign up for a free trial.

• Social 2013: The enterprise strikes back
• Social third-quarter 2012: analysis and outlook
• Readers weigh in: future prospects for Twitter

Draw Something, the mobile game sensation that put OMGPOP on the map, is getting new perks this week, including an easy undo feature, two-click posting of drawings to Facebook and Twitter, and messaging that lets users annotate their drawings before sharing them. And some of that work is already being done on Zynga’s internal Z cloud, Pearlman, who is now CTO of  Zynga’s new Mobile NY group, said in an interview Tuesday.

Z cloud is Zynga’s internally developed cloud that now runs the bulk of its computing workload. When Zynga started out, 80 percent of its game launch workloads ran on Amazon Web Services, now 20 percent of them do.

Moving fast is nothing new to OMGPOP. When Draw Something debuted in February, there were 30,000 initial downloads. Nine days later, downloads topped a million. To date, there have been 50 million.

“We thought we knew how to scale our backend systems, then Draw Something happened,” Pearlman said. “When we started with it, it was just another game to us. We make as many games as possible as fast as we can. We thought it had potential but had no idea how big it would get.”

It wasn’t as if OMGPOP went all out promoting the game. There was a small media buy, he said, but then people started tweeting about it. “Celebrities started tweeting and atop all that, the people who were signing up and playing weren’t leaving. Our DAUs [daily average users] kept growing and growing.”

Even engineers used to building to scale will tell you it’s hard to tell under these unanticipated loads “when things will start to fall over,” he said.

The OMGPOP back-end guys — Pearlman and two others — met the challenge with brute force. “We started bringing up tons and tons of servers, we ported over all our key value code to a new [non-blocking Ruby-based Goliath] framework and we worked around the clock,” he said. There were other hair-raising tales, about having to rewrite the app and deploy it on a new Couchbase NoSQL cluster and then hitting more issues.

Now Draw Something is up to 3,000 drawings per second. For that reason, Pearlman is really happy to be working with Zynga CTO of infrastructure Allan Leinwand on the strategy for moving more of OMGPOP over to Z cloud.

While there are benefits to working in small, tight-knit groups, there are also advantages to having lots of resources.  ”Having Zynga acquire us helps us — instead of spending all night just keeping things up and running,  we can work on new features,” he said.

Related research and analysis from GigaOM Pro:
Subscriber content. Sign up for a free trial.

• Social third-quarter 2012: analysis and outlook
• Takeaways from mobile’s second quarter
• Controversy, courtrooms and the cloud in Q1

The past one month has been interesting, to put it mildly. Facebook made an unprecedented move and snapped up Instagram for $1 billion. Earlier, Zynga snapped up OMGPop for $200 million (and change). The two deals have been put under the microscope by many (including myself.) However, here are three key lessons (and commonalities) between these two deals.

1. Panic & Act Fast:  Intel’s legendary CEO Andy Grove once wrote a book called Only The Paranoid Survive. It seems both Mark Zuckerberg and Mark Pincus read that book cover to cover. The speed with which Facebook and Zynga bought these two companies tells me that both Zuckerberg and Pincus are extremely paranoid about letting any competitor get big enough to threaten their business. The $50 million  infusion was all that Instagram needed to hire more people, grow its footprint and gain further momentum, and then Zuckerberg swooped in. The decisions to buy both these companies at such rapid speed tell me one more thing – the two founder/CEOs are firmly in control of their companies, boards and self aware of their weaknesses. And to me that is a good thing.
2. Engagement matters: Both Marks realize that the future of their business is not just the sheer size of their networks but engagement. The post-social world is an “attention economy.” If you don’t have engagement, you don’t have attention and if you don’t have attention – well you don’t have anything really.
3. All mobile all the time: In an interview earlier this year, Mark Pincus was pretty explicit in his desire to conquer the mobile platform. He was also very candid that none of the social web companies have figured out the mobile and this included his company and Facebook. So if they have to buy their way into the market, so be it. Facebook is no different. The company’s mobile efforts are a work in progress. Instagram figured it all out. That said, these two deals are a big indictment of Facebook/Zynga’s mobile teams: it seems the two Marks are also saying, we aren’t really confident about what we have in-house.

On Facebook-Instagram deal, Nabeel Hyatt makes a point that all founders should read.

The best way to think about the value of Instagram is not to think about $1 billion or $30/user, it’s this: 1%. There is no market more strategically valuable to Facebook than mobile, and there is almost no product more valuable than photos for them. It was a very unique situation where a $100b company made a 1% bet.

Related research and analysis from GigaOM Pro:
Subscriber content. Sign up for a free trial.

• Social third-quarter 2012: analysis and outlook
• Readers weigh in: future prospects for Twitter
• Takeaways from mobile’s second quarter

Now, it might not sound that interesting that Draw Something is doing well in China when many other top western apps are also hits there. According to App Annie, the free app was the 9th ranked iPhone game in the App Store in China and also the 9th-ranked paid game. Among iPad apps, Draw Something is ranked  for 7th for paid and 12th for free games. Infinity Blade, Angry Birds Space and Fruit Ninja are also top ten apps in China.

But stepping back, you realize this is for a game that is based on drawing English words and phrases, some of them slang terms. Almost all of the other top western hits on the Chinese charts are games, like Angry Birds, that need little localization support or can stand alone. But Draw Something is a huge hit (probably more than 50 million downloads worldwide at this point) even despite its language barrier in China and lack of localization.

Consider the fate of Zynga’s other top With Friends games in China: Scramble with Friends, Words with Friends or Hanging with Friends do not appear on the top 300 paid or free iPhone or iPad charts. This might help explain why Zynga was willing to pay about $180 million for the company.

China’s booming app download market

China’s share of downloads has grown considerably, according to app analytics firm Distimo. From early 2011 to November last year, China’s share of combined iPhone app downloads between it and the U.S. went from 18 percent to 30 percent in the course of 10 months. And China recently surged past the U.S. in the number of free iPad app downloads. That’s key for companies like Zynga that make their money more from in-app transactions and advertising in free apps.

I reached out to Dan Porter, the CEO of OMGPOP, to ask what he thought of the Chinese support for the game. He agrees it’s pretty surprising considering the game has had no marketing in China and is only available in English. He said that at one point last week it hit No. 1 in the App Store.

“It’s so crazy that something like this is so far away in another language and they’re still finding a way to use it. That’s part of the way it grows, it’s enough of a blank slate, that people can do that,” Porter said.

Getting creative with pictures

I came across an image of some Draw Something pictures out of Hong Kong, and they showed that despite the use of English words, the Chinese users were drawing pictures that would get recognized by fellow Chinese players. For example, for the word “Fox,” a drawer drew an image of a woman recently in the news for having an affair with a famous wealthy man. The slang for home wrecker in Chinese is “fox” spirit or demon.

Porter can’t talk about future plans for tapping the China opportunity or just how many downloads there are in China. The game is built for English and with China’s character-based language, it would be harder to translate the game’s word guessing feature. But it could happen through Pinyin, a system of spelling Chinese words with English letters.

For now, it’s a reminder of just how big the opportunity is in China and also how Draw Something’s simple approach for a drawing game can appeal to a wide audience. The game is less about competition but more about collaboration and communication. And with its open-ended style, people can use it for creatively expressing all kinds of words.

“It’s mystifying really,” said Porter. “But the way people have used the local slang and converted it into the game, that’s interesting.”

Related research and analysis from GigaOM Pro:
Subscriber content. Sign up for a free trial.

• Where the next-generation console fits in today’s video game market
• Social third-quarter 2012: analysis and outlook
• Takeaways from mobile’s second quarter

I had a brief call on Thursday with Couchbase CEO Bob Wiederhold, whose company worked with OMGPOP to scale its implementation of the Couchbase database as demand started growing. Although the companies aren’t ready to give exact details yet, here’s what Wiederhold told me:

• OMGPOP is hosted in the cloud, but “they’re not on Amazon.”
• Draw Something has been downloaded more than 35 million times. Players have created more than 1 billion pictures and are creating around 3,000 pictures per second.
• To handle the incredible traffic spike, OMGPOP had to reconfigure its Couchbase cluster, scale it into the many tens of nodes, and many terabytes of data and increased throughout into the tens of thousands of operations per second.
• Throughout all this, Draw Something didn’t experience any downtime.

This type of load really stresses a system, Wiederhold said, and if it wasn’t for its decision to use cloud computing and NoSQL technologies, “their game would have fallen over.” EA recently removed its “The Simpsons: Tapped Out” game from Apple’s App Store after server problems prevented users from being able to login. It’s not clear what, exactly, caused EA’s problem, but it speaks to the importance of having components that are able to scale as apps go viral.

Scalability, of course, is one of the primary calling cards for both cloud computing and NoSQL providers. NoSQL databases, which broke onto the scene a few years ago by claiming to solve the scaling problems inherent in many relational databases, are hugely popular among those building web applications. One of the early poster children of cloud computing was Animoto, who launched its Facebook app in 2008 and scaled to about 250,000 members and about 3,400 Amazon Web Services compute images over the course of a week.

Related research and analysis from GigaOM Pro:
Subscriber content. Sign up for a free trial.

• How data warehousing is now a cost-effective solution for businesses
• How direct-access solutions can speed up cloud adoption
• The promise of SDNs in the enterprise