The Techno-Optimist Battle with Reality
A response to Andreesen & Klein alike
In the tech world last week noted VC Marc Andreesen caused a stir with his publication of the techno-optimist manifesto, essentially a list of hundreds of twitter-length snippets smooshed together to make up a few thousand word laundry list of everything Marc thinks is good. The section headers make up the thrust of what techno-optimists are supposed to like, from clean energy and improved AI to more esoteric subjects like the politics of ‘abundance’ and the Meaning of Life (TLDR technology helps you figure it out better, Marc sure doesn’t know).
It’s a half-rant half-grab bag of current in-vogue discussion points around technology, and has gotten critical reception, put best by Ezra Klein’s note on Threads that “there is no starker proof…that the ‘medium is the message’ thesis than the way that the medium of Twitter has colonized the way Marc Andreesen thinks and expresses his thoughts.” It really does read like a twitter thread. The funny thing, as Ezra notes later in the post, is that much of the manifesto mentions the control humans have over technology, always being it’s master while it quite literally changed Andreesens own writing. Compare this manifesto to the last one, ‘It’s time to build’, written in early 2020 in short but clear paragraphs. The other half of the argument, that technology itself only provides value was challenged Monday morning by Matt Yglesias on the merits of innovation and technical advancement in the fields of fentanyl production & distribution, nuclear fission creating more dangerous weapons, and junk parts of the internet ravaging our attention.
I’m reminded of an author who spoke at my college inauguration my freshman year, Adam Gopnik, who had recently written about great men of history, and was asked to speak. It was a strange talk, I don’t recall much, but a few years later he was on the Ezra Klein show making a similar argument and it all came flooding back. In both cases, Gopnik who acted as the staunch defender of liberalism, made the case that liberalism, like techno-optimism, is a “practice before it is an ideology, a temperament and a tone and a way of managing the world more than a fixed set of beliefs.” It’s a grab bag of things that clearly work, like studies that found that expanding the Child Tax Credit during COVID-19 raised 2.9 million children out of poverty, and letting it expire caused 3.7 million children to enter or re-enter poverty as of January 2022. Never mind having a coherent idea of what a childhood should be like as long as the math works out.
The distinct lack of foundation is what both techno-optimists and modern liberal critics of those visions share. The lack of underpinning ideology allows both classes to wear ideologies like clothing, a critique leading to the modern use of the term “T-Shirt Liberalism,” about how people can work themselves into irreconcilable knots of belief versus action, like a neighbor who has ‘we welcome everyone’ yard signs but shouts down new development in local planning meetings.
Regardless of the political coherence issues plaguing liberalism, its lessons about how those issues lead to a cycle of entrenchment must not happen to those of us who call ourselves Techno-Optimists. Marc Andreesen’s harm in his manifesto was around this paradoxical rootless entrenchment. Rather than ideals for a strong future the writing felt more like a list of current grievances & perpetrators, sinners & saints.
To try to make amends, rather than write a brooding manifesto dripping with my own self-importance, here are five key relationships around technology that I believe will shape the next fifty-odd years. I am personally optimistic about each, and so they read with only a touch of self-important belief. I hope you’ll forgive me:
Build & Partner for Personalization – The data world is moving towards a paradigm of first-party preeminence, as this Substack has repeatedly written about. From digital data laws, privacy standards, data acquisition laws, and more the internet business model of the early 2010s where you could buy any data you wanted is nearly dead & gone. Google has almost fully deprecated web cookies (anytime now…), and partnerships between companies with richer datasets is providing far more business value than paying $2M for a dataset that gives you $2.1M in returns (if you’re lucky) in the Ads markets. The downside is that this is, ironically, a far less transparent world for the consumer. Who knows how your data is used once it’s in Disney’s hands. Before, it was just company policy & consumer concern that prevented wide scale data sharing. Now that sharing in ‘private exchanges’ is the gold standard, what happens behind closed doors to your data?
Government enters the stage – President Biden’s PMA, aka his goals for the government, ends with a commitment to ‘improve our internal processes and modernize technology.’ In China, while wiping out over a trillion dollars of commercial technology, Xi has ramped up OSINT collection staff by the thousands, pulling that capability into the government. Russia, in its war with Ukraine, has expanded its disinformation capabilities to the max as its ground war falters. And on & on. The governments of the world are coming into the digital game in a big way. The big fish mostly hunt each other, but we will be caught up in the damage.
AI will be more specialized, not more general – ChatGPT and other generative AI systems have caught the curiosity of the world at large, sparking debates around fully functioning AGI which led to more conversations to ban development on AI. However the rapid follow-on advancements that are getting taken up by the industry as a result of LLM and generative technology are less general and more industry-specific, as has been the case with previous rounds of AI. DeepMinds work into healthcare diagnosis support or Viz.ai with medical imagery. DuoLingo into education personalization, various groups using it for climate carbon recapture, and so on. There’s been almost no progress on the issue of consciousness, but that’s not what modern technology is meant to solve. Personally, wake me up when there’s something more interesting than matrix multiplication, until then I’ll take industry-revolutionizing innovations.
Somebody will figure out security – To be blunt, the world of cybersecurity is a fucking mess. Look at this public blue team handbook table of contents (blue teams try to secure cyber systems and test for vulnerabilities). The device, the OS, the user, the browser, the network, the software, the patch to the software, the employees' kids or spouse, or their own version of all of the above assessed incorrectly opens a million doors for vulnerability. How many times do I have to take a company-sponsored ‘don't do anything on your machine ever’ before we realize that it’s not working. Phishing, Malware, Password attacks, unsecure website attacks, and DDoS make up almost 90%+ of all attacks. Google has ‘defense-in-depth’ meaning you just have to protect everything correctly. Microsoft says ‘Security is everyone's responsibility.’ IBM says ‘Security is a journey, not a destination.’ Sounds like a lot of cope to me. Google faced the largest DDoS attack in history, again, last week. Despite executive orders, most of our government is not even on a zero-trust standard. Time to get cracking as we’re already getting regularly hacked to oblivion anyway.
Bits & Atoms – As the information economy expands, getting technical solutions to ‘the edge’ is getting easier every year. At Google we worked on video intelligence AI to help ID wildfire damage in California, worked with Memphis to detect potholes and schedule repair, tested poop for upstream COVID-19 case prediction, and on & on. As we expand the Alexandrian Library of ML models, innovation will come from combinations unforeseen even by the model builders with physical tech. The recent LLM craze powering ChatGPT, for example, is being used for fraud detection in financial systems to help ID suspicious activity that even a well trained analyst would be hard-pressed to discover.
Each of these predictions is a result of, and further step expansion towards issues with scale. Personalization & data-first partnerships are only possible today with the scaled possibility of first-party data management generally provided by Cloud technology, which is also true of specialized AI and Cybersecurity. The public sectors of the world's largest governments entering the digital arena moves mountains in a way no other entities can, but also helps drive the collision of Bits & Atoms that takes more zeros than VC money can provide to jumpstart projects.
We can see an optimistic future without idolizing technology – governments get better at clamping down on cybercrimes, new gadgets with smart tech continue to drive an already-greatly-improved economy since COVID, and personalized services only improve quality of life. However, if governments engage in increased disinformation as a ‘defense’ against opponents overseas, if smart tech pumps out bullshit like the crypto/NFT wave, and if personalized data only decreases consumer choice when paired with increased consolidation of industry, then that future is not bright at all.
Like with every past wave of technology, this is no different. These are just tools, and what matters is how we use them. Today, the impacts are just higher when innovations are scaled to billions seemingly instantly. Place your bets now.