Betaworks Trilogy, Part III: Thinking What's Next
If someone forwarded this to you, you can subscribe here or via the button below.
You’re reading Part III of a deep dive into the origins, work, and ethos of betaworks.
In Part I, we went back to the firm’s foundational beliefs and building roots.
In Part II, we unpacked how betaworks grew into the multifaceted business its founders always envisioned.
For the third and last part of this series, we’re exploring how betaworks’s ethos has prompted it to build a better internet for all — and which technologies the firm is now leveraging as it enters its next phase.
Thinking tech through
By now, we've covered betaworks' founding principles and strategy. But to understand what the company is really about, we also need to dig deeper into its ethos.
At the heart of it is an ability and a willingness to think tech through — that is, to consider the many ways in which a particular piece of technology could be misused, either intentionally or unintentionally. This, too, seems to stem from Borthwick’s personal ethos, which he eloquently touched on many years ago:
I think the line between computing, the interface, data, and human thought, is a lot grayer than we acknowledge normally. I think we think of this as something else, and it's not, it's us. It's part of what we've created. To me that's hugely important to believe, because it makes us accountable for what we've created, and the follow-on effects of it. Because I think technology has a grain to it, I think it has directionality to it, and I think there are unintended consequences to things we build. And as builders, we need to own those responsibilities and live to them and see if we can make better worlds through them.
This approach has lived on. Through both their words and their actions, the team makes sure to consistently point out tech's potential dangers so as to subdue them before they go rogue. Here are three examples when the firm called the public's attention to a looming issue.
2015: Media Hacking
Back in 2015 already, shortly after "fake news" had entered the mainstream lexicon and become a tangible threat to democracy, Borthwick published a piece titled Media Hacking. Pointing to two instances where false information had spread inconspicuously across social networks, he made the case for how better discovery, measurement, and attribution tools could enable us to track and assess information propagation.
These tools, of course, are more needed than ever. The speed at which misinformation circulates has only increased as social platforms continue to grow, social graphs get richer, and we get more accustomed to more media — on TikTok, for example, video-based misinformation lives almost unconstrained. Borthwick's analysis was a word of warning.
2019: Synthetic Media
In July 2019, then-Betaworks member Sarah McBride posted a piece titled Investing in Synthetic Media. Its first sentence: "How “we” at betaworks are thinking about this emerging market." There was just one problem: the entire post was written by an algorithm, as revealed one week later by Borthwick.
It was made "using a transformer-based language model (gpt2)" and feeding it with the team's first post on the topic, written 7 months prior. Betaworks described it as "an experiment in synthetic media," noting:
“We” found this experiment instructive — all forms of media, text, audio, images and yes videos are colliding with machine learning requiring whole new systems and frameworks for trust. That is what we actually do want to invest in.
Like misinformation, the capabilities of synthetic media have increased tenfold since this experiment took place. Synthetic faces and voices are now commonplace and will increasingly put our media literacy to the test. I like to think that betaworks's early call for "frameworks for trust'' was heard by the first wave of voice synthesis companies, who together have shown strong ethical commitment.
2020: Fixing the Internet
In March 2020, the team launched Betalab, an initiative that aims to "fund startups and researchers focusing on how to fix the internet from data and privacy to attention hacking and misinformation."
Part of the Camp model, Betalab combines "an early-stage cohort-based investment program" with "a year-long series of workshops and events." At launch, it shared a fascinating list of "Requests for Startups," across categories that spanned:
"Identity and authentication tools;"
"Software that builds trustworthiness into AI;" and
Later in July, betaworks announced that Yaël Eisenstat, a strategist working at the intersection of ethics, tech, and security, was joining the program as a Researcher in Residence. Eisenstat's role there is to ensure that
"the decisions companies make as they grow are grounded in an understanding of the societal implications of new technology."
Of course, not everything betaworks tinkers with poses the same level of threat. It's unlikely that Leo AR, an app that lets you create AR videos, could seriously harm democracy.
Yet malign consequences can indeed come from seemingly benign products. Betaworks's GPT-2-powered article on Synthetic Media, which fooled many a reader, was produced using a deceivingly simple web app. The most fun of face filters can resurface concerns over foreign data mining. And while custom GIFs are fun, the same deepfake technology is already eroding what little trust we have left in the news media. Clearly, we shouldn't judge a book — or a GIF — by its cover.
Betaworks's Red Queen's race
It's notable that one of Betalab's areas of focus is "Software that builds trustworthiness into AI, protecting, detecting, or authenticating synthetic media or data." Through its own investments in companies like Morphin, Facemoji, or Resemble, betaworks arguably played an active role in advancing the technologies it now wants to tame.
But it would be hypocritical to expect the firm to refrain from betting on these technologies in the first place.
We wouldn't have the internet if we'd stopped building it at the sole prospect of fake news and deepfakes. Nor would we have YouTube if someone had warned us of political radicalization or gore kids' content. Technological innovation and fear mongering often go hand in hand as pessimists try to hamper what they intuitively see as a threat. Today's media misuses and attention hacks are the price we pay for accessing what has become vital infrastructure. What matters is that we do see these issues for what they are.
In a way, betaworks is perpetuating its own Red Queen's race, alternatively encouraging technological advancements and preventing their weaponization. That the team is simultaneously excited for and wary of deepfakes is proof that a single technology can legitimately inspire ambivalent feelings — and call for similarly ambivalent treatments.
Discussing our collective longing for truth through technology, Lux Capital's Josh Wolfe aptly described a never-ending struggle between deception and detection, an "arms race for reality." Likewise, betaworks's technological bets aren't competing as much as they are complementing each other: deepfake generation and deepfake detection are two sides of the same technology that we know for certain will need to coexist from now on. As an investor, the only moral option, if you're backing one side, is to back the other, too, and betaworks has repeatedly risen up to its responsibilities in that respect.
When I started working on this piece late last year (I’m a slow writer), betaworks seemed to be going through a new period of reflection. AudioCamp, the latest edition at the time, had closed in April 2020, and the firm hadn’t publicly set its sights on any particular technology yet. This forced me to speculate a bit, suggesting it might be time for the company to run its first gaming-focused program, or a new and improved VisionCamp that would build on top of AR platforms like Niantic’s Lightship.
The former, in particular, didn’t seem too far-stretched, considering gaming has been on the team’s mind for some time. Half-hidden in betaworks' portfolio of products is Katapult, a digital entertainment company building CHKN, an open-world sandbox game. Additionally, Peter Rojas, previously a partner at the firm, had been prominently using a match-making service, Matchbox.vc, to connect with founders over games, and started sharing his thoughts on social gaming before leaving to join Meta.
While neither of my two predictions materialized, there is no need to speculate any longer. The company itself has laid out its plans for us: its focus, its website states, is now “on web3, crypto media, NLP/CV, applied machine learning and tools for thinking.” Here, I’ll focus on two of these themes and how they relate to betaworks’s history and interests.
Tools for Thinking
In November 2021, the company debuted Nextcode, a series of intimate virtual events dedicated to no-code tools and the experiences they enable, including "GPT3, games that let you create games, & productivity workflows." Enough came out of it that betaworks saw the need for Thinkcamp, a recently announced program dedicated to “tools for thinking,” that is, “tools to relate, organize, scaffold and add depth to the [information] landscape so we can individually and collectively make sense of it.”
Though this may seem like a departure from betaworks’s historical focus on consumer media, it’s anything but. By “tools for thinking,” betaworks means not just the “productivity tools” of old, but indeed a new category of tools now shaping how we think. Like the internet, the home screen, and the notification layer, like bots and voice interfaces, these tools are a sign of the times: an opportunity space brought about by a particular combination of new technologies (e.g. “graph databases and Web3,” the announcement mentions), new interfaces (“dedicated hardware, next-gen web browsers, voice and AR”) and “a massive step function in enabling technologies” (“NLP and LLMs,” respectively Natural Language Processing and Large Language Models).
There’s another reason that makes this field especially compelling: these tools are about networks. Networks of thoughts, conceived and presented by companies like Roam and Obsidian as atomic units all interconnected within ever-spreading knowledge graphs. Networks of thinkers, now able to share said graphs with others to let the world build upon specialized bases (e.g., Startupy) and ready-made trains of thoughts. For a company obsessed with unlocking the potential of the collective, this is well worth exploring.
Web3 has certainly been on everyone’s mind these past few months. A massive influx of both talent and capital has converged to hundreds of startups leveraging the potential of blockchain technology across a myriad of use cases. As an investor and builder located in one of the world’s hottest crypto spots, it makes sense that betaworks would get involved with the technology and its fast-growing ecosystem.
Yet I think this interest isn’t just opportunistic. Rather, Web3’s ethos and capabilities align with betaworks’s vision perhaps most closely that they do most investors’. Remember the company’s original assumptions? If we try to map them to what we know of blockchain technology, they happen to match almost perfectly. Let’s see how.
“Each new platform shift introduces fundamentally novel ways for consumers to interact with technology.”
Blockchain technology as a whole comes with a number of unique traits, each of which is powering interactions previously not possible. The combination of immutability and verifiability, for example, lets users readily search and pull data on any particular asset using tools like Etherscan. Trustlessness means you can expect a smart contract to run without fail, a prerequisite for automated royalty splits, while composability enables protocols and applications to leverage each other’s capabilities instead of reinventing the wheel.
This premise remains true if we consider more specific aspects of the user experience. Wallet-based login, for example, replaces legacy site-specific methods of authentication based on a user's email or socials. As Seyi Taylor aptly puts it, the wallet simultaneously represents your identity, carries value, provides access, and preserves memorabilia. Through it, the circulation and utilization of a user’s entire online history, tastes, habits, and relationships is just a click away.
“Social was fundamentally changing the way people found, navigated, and shared content online.”
Simply replace “social” with “blockchain” and the assessment is relevant afresh. By leveraging wallet-based identity, a wave of startups have started to build new experiences around content discovery (e.g., Context) and curation (e.g., Cyber). The Worm — “a Share-to-Mint NFT” — uses wallet-to-wallet transfers to grow longer, leaving a copy of itself in every wallet it visits. Creators and communities alike can use granular token-gating (e.g. Guild.xyz) to unlock exclusive content, access, roles, and perks. Finally, mechanics like airdrops and splits help automate the dissemination of cultural and financial value. Across all these use cases, entirely new social graphs are being built.
“A loosely coupled network of companies, connected at the data level, could hope to compete with tech giants and their accumulated advantages.”
What better example of a transversal data layer than the blockchain? Consider Ethereum, which facilitates the recording, security, and connectivity of all data on the network and ensures that it can be accessed, read, and used by the thousands of protocols now building on top of it. Although these protocols may each be working on very different problems, they all have access to the same, ever-growing ledger — “a loosely coupled network” indeed. From there, each of them is able to query transactional data, for example to turn it into quantitative insights (e.g., Dune) or showcase provenance (e.g., JPG).
When it comes to the second part of the assessment, the jury’s still out. On the one hand, the ability to readily tap users’ history enables Web3 players to quickly come up with unique concepts and features. On the other hand, it remains to be seen how much of a threat to incumbents these companies can collectively pose. DeFi overall has proven significantly more capital-efficient than many legacy players, and a company like Pollen is similarly disrupting the capital-intensive process of 5G deployment. On the consumer social front, however, usage on most services undoubtedly pales in comparison to any of the Web2 leaders. Competition, at least for now, has more to do with users’ adherence to what they deem a new, more compelling ethos — of decentralization, openness… — than it does with DAU count.
“A growing technological infrastructure was making it increasingly faster and cheaper to create and iterate on companies.”
Among the blockchain’s most discussed traits is composability — the ability for developers to create new smart contracts by “remixing” components of existing ones, thanks to open-source code. This enables applications and protocols to build off one another’s capabilities and come up with faster, more capital-efficient solutions than currently exist. When combined with forking — “the process of creating a new version of a system by copying it under new ownership” —, composability fosters fast, ecosystem-wide innovation by increasing competition. It pushes individual players to serve users the best they can, most often through financial incentives used to kickstart initial adoption and liquidity.
All this has prompted the emergence of a rapidly growing infrastructure that newcomers can leverage to spin out a new app or to add specific functionalities inside existing products. In gaming, for example, developers can tap highly-specific building blocks to issue tokens, launch an internal P2P marketplace, handle players’ funds and identity, and more. This lets them focus on what they do best and allocate resources to areas like game design where they are most likely to differentiate from their peers. Ultimately, this indeed means faster, cheaper development, go-to-market, and iteration.
Overall, I think Web3 perfectly embodies betaworks’s mindset and goals. At this time, the firm has already made public several investments in this space, including: XMTP, a wallet-to-wallet communication protocol; Unlock, a token-gating protocol for creators and communities; Infinite Objects, which develops consumer hardware for showcasing NFT art; Ink, which lets artists digitally sign any digital content and sell it as an affordable NFT; and Medallion, an artist-to-fan platform. (Superplastic could also be added to the list, although it didn’t start with Web3 in mind). In addition, some light Twitter sleuthing shows that the firm may be somewhat involved with both Deca and Tableland. Except for the more infrastructure-leaning XMTP, all these deals are representative of what the team calls “crypto media,” that is, crypto-enabled creative content and the experiences being built around it. No doubt that exciting products will emerge from this new playground.
With its finger always on the pulse of consumer media, betaworks has been able to make the most of the industry’s never-ending stream of innovations. Over the years, the team’s vision, product acumen, and multidisciplinary network have made it a trusted provider to tech giants like Twitter, Meta, and others.
But times are changing. Today, the acquirers of old are under growing scrutiny as they look outward for innovations they could integrate into their rosters. Meta is facing legal action from the FTC, which seeks to block the company from buying Within, the maker of VR fitness app “Supernatural.” Similar hurdles await on the other side of the Atlantic, where the UK’s Competition and Markets Authority has been trying to block the company’s acquisition of Giphy for more than a year. As regulatory concerns over these giants’ dominance grow, the traditional path to exit may be getting bumpier.
Should betaworks be worried? Probably not. For one thing, there are, and will be, other buyers: companies like Unity, Epic Games, Roblox, and Snapchat, among others, have not yet been scared out of the M&A road. Should that door shut too, betaworks can afford to wait. In any case, it will be here, doing what it does best — building.
Many thanks to John Borthwick, Matt Hartman, Billy Chasen, and Andy Weissman for reviewing early drafts of this piece and sharing their insights with me.
Thanks also to Devon, Fawzi, Neer, Pénélope, and Morgan for their thoughtful feedback.
Thanks for reading. As always, I’d love to hear your thoughts! You can find me on Twitter, or just reply to this email.
If you’ve enjoyed this essay, please take a minute to share it with others.