In “The New Virtual Accelerator”, we posited that all of venture can be broken down to knowledge, networks and capital. Today we focus on the first leg of the stool, knowledge. The problem is not that there is too little startup information publicly available–quite the contrary. The big problem regarding online resources are discoverability, quality control, maintenance, and to a lesser extent, access. This piece will not focus on those pieces of knowledge that you rent (e.g., legal services), buy (e.g., hiring an in-house lawyer), or gain via connections (mentors), but rather that you discover and learn on your own.
I’ll wager that within 3-4 years, most of these knowledge problems listed will be resolved, and that one comprehensive, highly vetted online solution will emerge as the overwhelming single source for search and distribution of high quality startup content. While no such utility exists now, let’s ponder how it should be designed and built. If you want to share your ideas or help build it, ask for an invite to the Ultimate Startup Knowledge Base Facebook group.
The slogan “Information Wants to Be Free” was coined by Stewart Brand and embraced by technology activists well before even the primordial days of an “information superhighway”. But as Andy Weissman points out, what Brand really meant by “free” was not cost (free = “gratis”) , but access (free = “libre”).
Lots of content now is free in both senses. Early VC bloggers like Brad Feld made their reputations by demystifying the process and creating educational content. Simultaneously, online education from EdX to Khan Academy took off. Today, educational content of various quality is scattered everywhere. So where to go for startup-specific content, and how can you identify the best content amongst all the mediocre sources?
That one gigantic knowledge aggregator will emerge from the pack seems pretty inevitable: Ben Thompson of Stratechery points out that as the cost of moving information goes to zero via the internet, aggregation theory comes into play, with a few big winners (e.g., Google, Facebook, Amazon) ending up with the lion’s share in their field.
The Minimum Viable Product: Full of Info, but Hard to Navigate
Like most standard aggregators, from law (WestLaw) to medicine (PubMed) to Google itself, the Ultimate Startup Knowledge Base (let’s call it “USKB” from here out) will emerge from an MVP consisting of a vast array of links tagged to certain topics. Many universities, venture capital funds, law firms and accelerators and others already have put together and sometimes published their own selections of resources; v1.0 of the USKB will be no different. It will be “better” in that it can be more comprehensive– Stanford doesn’t necessarily include all of Harvard’s resources, and vice versa. A site that lists both together has more potential utility.
Finding good starting material for the MVP can be done relatively quickly. The top VC blogs/slideshares/podcasts such as Fred Wilson’s avc.com; VentureHacks, AskTheVC; material from top business schools (Stanford, Harvard, MIT, etc.) all make for a good base, and could be catalogued in a summer by one dedicated individual without the need to build wikis, editorial panels, etc. Think “Jerry’s Guide to the World Wide Web” before it became Yahoo. So, the first step is simply building a base reference that can be expanded upon. No need to get fancy with machine learning…although I do have a soft spot in my heart for the hilarious Paul Graham Essay Generator.
The Discovery Problem
Plenty of content is out there already. Do a Google search on anything related to startups–there are literally millions of results, supposedly stacked by smart algorithms that bring the most relevant content to you first. But does it really? Here are 9,650,000 results for “best startup information blogs”.
Assuming Google’s ranking algorithm is effective regarding startup knowledge is like assuming that the number of meals eaten is directly correlated to the quality of restaurants. (Which only is true if you believe that a Big Mac is the highest achievement of cuisine.) The top results often are listicles optimized for SEO, not quality. What we need to do is discover the most relevant high quality content, not merely the most popular.
Sourcing Content and Quality Control
While there can be a case made for the wisdom of crowds, and machine learning may be able to make sizeable dents going forward, I believe the USKB is best served by a wiki approach managed and edited by experts. Think of PubMed, which is used by medical researchers and academic clinicians to find the primary research done on medical topics. The articles are all sourced from indexed (i.e., of sufficient quality) journals headed by highly regarded experts in the discipline. Similarly, I believe that the database will be initially populated by articles selected by some team of expert editors from known, highly regarded sources. Thus, the experts solicited to edit any articles pertaining to Software as a Service might begin with content published by recognized mavens (e.g., David Skok, Tom Tunguz, Jason Lemkin). N.B.: just because someone is recognized for creating great content hardly means that most of their content is suitable. Few people if anyone has a record of producing better content than Fred Wilson, yet over the years his archives show far more written about his favorite music (893 posts) and photos (178) than topics from general (management–63) to specific (blockchain–49). In other words, no matter the author, you have to go through a lot of chaff to get to the wheat.
Navigation: To Search or to be Led?
Search is great if you know precisely what you are looking for and need–the “known unknowns”. But as Donald Rumsfeld points out, what about the “unknown unknowns”? In those instances, you need a guide.
The USKB Ultimate Startup Knowledge Base needs to be able to offer solutions to all the “unknowns.” The latest and greatest tech tools can’t fill your needs (“Alexa, play my favorite music”) if it doesn’t know anything about you. So in addition to standard keyword search for “knowns” (be that by topic, author/source, semantic), the USKB needs to figure out where you are on the startup map, so to speak. Are you a beginning entrepreneur without cofounder? A funded team with a product and revenues? Both may pose the same question, but require different answers. Do properly parse the distinctions requires a profile and the ability to do a diagnosis.
The User Profile
If a health insurer can give you a quote based on a few questions, so should the USKB be able to better assess your needs with more knowledge of who you are, your existing knowledge and even your usage habits. (Never watch a video more than 30 seconds through? Stop recommending videos…)
The USKB needs to be tuned to the individual using it. In addition to keeping a (private) profile of the individual using it, it will be able to improve results when given additional information on how the knowledge is meant to be used. Thus, profile questions of the circumstances (e.g., work done within a particular company) are needed as well–there’s no need unearthing the best information ever assembled on scaling a team from 100 to 1000 engineers if the real question is not “team building”, but “how can I find a technical co-founder?”
For those demanding anonymity, standardized profiles can be used for either the individual or the company. For those not demanding privacy, in some future iterations of the USKB, I can see being able to make your profile searchable/accessible, just as it can be done now via LinkedIn, GitHub, or AngelList. Companies can be profiled not just for size/revenues, but also tech stack. (Companies recruiting on AngelList do this now.)
It could be that the killer way for the USKB to take off would be having a recruitment section attached–companies could not only add on MOOCs they make for their employee training available more broadly, but provide testing of job applicants for skills and fluency about concepts. Great course content => more eyeballs of potential hires. If that is the case, the USKB could become a must-have piece of the puzzle for large tech companies, which would not only provide content but perhaps sponsorship. On the other side, job hunters would be able to demonstrate their thirst for ongoing education and their interests by making their usage stats and personal information public.
The Recommendation Engine
The combination of a personal profile with context allows for the USKB to pinpoint exactly where you are on the map and suggest appropriate directions. Let’s say someone is considering learning about growth hacking. That discipline lies at the center of a Venn diagram with connecting marketing, programming and statistics. Individuals coming from one of those three disciplines, once their profile is known, might see material from the two other circles, while a complete newbie will get exposed to all three of the core elements.
Just as Pandora used the “Music Genome Project” to analyze songs by 450 qualities and then tailor recommendations, so can the USKB predict suitability for the profiled reader. Then, as the profiled user gives thumbsup/thumbsdown feedback on the suitability of the recommendation, the recommendation engine learns and improves.
MOOCs and the Prescribed Curriculum Approach
The academic community already has open-sourced many of their course offerings. Stanford offers lectures (“How to Start a Startup”) and multi-media resources (https://ecorner.stanford.edu/); Harvard and MIT post some of their content via EdX for nominal fees, with some professors simply open-sourcing their classes (e.g., StartupSecrets) or years’ worth of classes, reading and topic lists. The curriculum approach, while static, makes it easy for someone to pursue a new topic in an orderly and considered way, with degree of difficulty taken into account in the material selected. I would anticipate professors creating their own class curriculum—perhaps even in a for-profit Massive Online Open Course (“MOOC”)—from the resources available, which will be constantly upgraded and kept topical.
Other Discovery Formats: Grids/Network Maps/Controversial Topics/Crowd-Ranking
Two years ago, I taught a class at Middlebury College called StartupGrid. Students created essentially a giant spreadsheet, with people (generally authors) in the columns, and then the topics in the rows. At a glance, they could see and click through what an author had published or peruse a topic and survey the different takes on the subject. This approach could be upgraded into network maps, and be a different or cool way to discover topics and content. What I found most interesting, however, was discovering where there wasn’t consensus–that led to the most interesting thought and learning.
As an example, take founder equity. The unanimous expert consensus is that a multi-year vesting schedule is necessary. (Thus, you don’t need 100 articles listed on that topic, just the best ones.) However, what’s the best way to split founder equity? Now THAT is controversial. Right down the middle in even shares, as per YC? According to Joel Spolsky’s Fairness Formula, which guarantees different ratios? Dan Shapiro’s take, which disagrees with both? Keeping the majority with a primary founder, as per Mark Suster? I’d love to see a separate “Discover Controversial Opinions” section.
Thus, you can also create sections based on conflicting opinions and approaches.
And inevitably, there is crowd-ranking. “Most popular articles” might be a popular feature, although personally I would want to avoid this. Popularity doesn’t get you new and different, it gets you Justin Bieber.
Calling Dr. Watson and the Machines
Unfortunately, much material is hard to tag or classify, especially anything that is non-text based, long, and covering several topics. Imagine a filmed panel session where several different takes are presented by the speakers? What I’d like to see is IBM/Watson (or some peer service) volunteer to transcribe data from sessions, tag it appropriately and put time stamps on the appropriate sections. Then the recommendation engine could direct the user to the precise correct location. NB–we need a common, published taxonomy of terms.
How do you keep the USKB up-to-date, adding new materials and reordering those “best” answers? I’m guessing that this, like Wikipedia, becomes a full-time job for a few, with the bulk of the remaining non-computerized work being done by students in university entrepreneurship classes. Interested in how 25 Middlebury students catalogued 18,000 items in one month? Here’s their code on GitHub. Want to know more? Sign up on the USKB Facebook group.
Workshops: I can also see the USKB being a site for asynchronous workshops. Topics like “How Should Founders Divide Equity” or “Which Product Features Should I Develop First” lend themselves easily to videos meant for group viewing, exercises and discussion. Many accelerators and executive coaches conduct these types of participatory exercises for their companies. I’m confident that as the USKB picks up momentum, more and more solid, engaging (and branded) content will be created by professional groups and service providers.
Tools: There is no reason for entrepreneurs to ever be confused about cap tables again. Cap table calculators, metrics calculators, standard accounting packages, templates, open-sourced functions like Social Capital’s EightBall, whatever is useful and vetted.
AMAs and RFAs: This is a natural, especially to fill in the blanks. I tend to think that we should not just repurpose existing material here, but also commission it, the way that Frank Fabozzi got the world’s leading financial experts to write all the chapters of his books for him. Free labor, but he got the rewards. On SaaS, for instance, we could solicit all the people who have been on the SaaStr podcast, creating a USKB SaaS curriculum.
Standard “B-School” information: While we are discussing the Ultimate STARTUP Knowledge Base, I see this more as a technology-oriented business base, focusing on startup needs. But much of the knowledge needed pertains to all businesses, regardless of the state of maturity. There is no reason that standard business school management lessons can’t be included, including behavioral interviewing, zero-based budgeting, cost accounting, etc. Second-tier business schools should be very scared. When knowledge is readily available online, and without the networks and branding of the top b-schools, why should anyone bother attending? Ditto for second-tier accelerators.
Public vs. Private, Free vs. Paid
While the USKB is “free” in all senses–both in access and in price, the truly valuable information will remain closely held and unavailable to the USKB. Confidential, internal data (for instance, accelerators’ history of investor interactions with startups, investor rankings by their portfolio companies, off-the-record founder stories, private phone numbers…) are unlikely to be published, but rather used by the owners of that data as a proprietary weapon. But frankly, the large bulk of content from any of the major players is commodity-like in nature–and that will be public and free. But for the good stuff, the secret stuff–that comes at a price, or only to members of the club, be that the companies and entrepreneurs within that elite accelerator, university program, or VC portfolio.
Who Will Organize This?
My guess is that no current single venture player can pull this off–not any single VC, nor any single accelerator, nor search provider. It’s not a matter of resources–I think it will be fairly straightforward to attract sponsorship from the major vendors for tech (banks, law firms, venture capital firms, software giants, data storage/compute companies)–but rather that winning aggregator will need to harness the cooperation and goodwill of many, many players. I think that means the provider will be some type of consortium, non-profit, or new venture that becomes some type of common utility.
But for now, it’s up to us. Who’s in? If you’re interested, ask for an invite to the USKB Facebook Group here.
(Thanks to Rob Rubin of Microsoft and the Internet of Learning Consortium for his enlightening talks.)