Yet, some of the most successful consumer products started as what we’d now call tarpit ideas. Think Facebook, Instagram, DoorDash, Yelp. You could argue the problems they were tackling weren’t as broadly appealing at the time, but they weren’t first in their category either. There’s hindsight bias too — they were so wildly successful that we really can’t call them tarpit ideas anymore.
Crypto’s trends from the ICO boom; to NFT summer; to socialFi, to memecoining, show me that people like to do their own research, get some sense of market advantage and then buy in size.
The iPhone’s launch in 2007, was clearly disruptive to the existing “smartphone” market controlled by Blackberry and Palm Pilot. But a few years later, the iPhone went on to disrupt nearly every industry via the creation of the mobile interface. (And we predict that a similar delayed reaction is likely to happen in generative AI as well).
The end of 2023 marked an inflection point for the presence of memes on chain. Averaging about 9,000 new tokens per day from August to November of 2023, Solana now averages more than triple that at 28,000 per day. At its peak, it topped 100,000 new tokens per day using the 30-day moving average (670,000 outright).
One general rule of technical advancement is that it’s not necessarily the most feature rich variant of a new technology that reaches the tipping point and critical mass, or even the cheapest or most available: rather, it tends to be the easiest to use.
And as Sam Altman once said, every year we get closer to AGI everybody will gain +10 crazy points.
—
In any case, people overrate the importance of open-source as we get closer to AGI. Given cluster costs escalating to hundreds of billions, and key algorithmic secrets now being proprietary rather than published as they were a couple years ago, it’ll be 2-3 or so leading players, rather than some happy community of decentralized coders building AGI.
If the last couple waves of startups felt like 10x improvements, AI provides what feels like a 100x better experience than the incumbent substitute (humans!) by compressing what is almost always the significant effort of hiring and managing another person, into a near instant experience that will only get better over time. To do this at a small fraction of the cost of hiring/managing that human dramatically opens up limitless use cases and therefore dramatically expands the market. If people underestimated the size of Uber’s market initially, we’re all underestimating the size of many AI startups’ opportunity.
The analogy here is to Search, another service that requires astronomical investments in both technology and infrastructure; Apple has never built and will never need to build a competitive search engine, because it owns the devices on which search happens, and thus can charge Google for the privilege of making the best search engine the default on Apple devices. This is the advantage of owning the device layer, and it is such an advantageous position that Apple can derive billions of dollars of profit at essentially zero cost.
Coinbase: Last piece of big puzzle is international expansion. 17% of revenue today. We’ve picked 10 markets. “Go deep markets”
Though the team used to run a marketing agency, working with brands like Casper in order to fund MSCHF projects, they stopped taking on clients last year. Now, they pretty much do whatever they want.
—
“Everything is just, ‘How do we kind of make fun of what we’re observing?’” Mr. Whaley said. “Then we have as much fun with it as possible and see what happens.”
Ethereum is a vision with a business while Solana is a business with a vision.
On the one end there are the doomers. They have been obsessing over AGI for many years; I give them a lot of credit for their prescience. But their thinking has become ossified, untethered from the empirical realities of deep learning, their proposals naive and unworkable, and they fail to engage with the very real authoritarian threat.
In the early folds of the paper—for instance, when you’ve folded it seven times and it’s still less than an inch thick—it is hard to see how it is possible that on fold fifty, a thin piece of paper could reach the sun.
—
we get frustrated when our wifi won’t transfer in two seconds what would have taken twenty minutes in the year 2000.
The project that has done the most on the former is perhaps Worldcoin, of which I analyze an earlier version (among other protocols) at length here. Worldcoin uses AI models extensively at protocol level, to (i) convert iris scans into short “iris codes” that are easy to compare for similarity, and (ii) verify that the thing it’s scanning is actually a human being. The main defense that Worldcoin is relying on is the fact that it’s not letting anyone simply call into the AI model: rather, it’s using trusted hardware to ensure that the model only accepts inputs digitally signed by the orb’s camera.
Modularization incurs costs in the design and experience of using products that cannot be overcome, yet cannot be measured. Business buyers — and the analysts who study them — simply ignore them, but consumers don’t. Some consumers inherently know and value quality, look-and-feel, and attention to detail, and are willing to pay a premium that far exceeds the financial costs of being vertically integrated.
When hype hit Snapchat, the product and growth loops had been maturing for months without any distortion from hype.
Rather, a useful way to think about generative AI models is that they are extremely good at telling you what a good answer to a question like that would probably look like. There are some use-cases where ‘looks like a good answer’ is exactly what you want, and there are some where ‘roughly right’ is ‘precisely wrong’.
The number of Chinese websites is shrinking and posts are being removed and censored, stoking fears about what happens when history is erased. China’s internet had 3.9M websites in 2023, down ~27% from 2017; Chinese-language websites were 1.3% of the global total, down 70% from 4.3% in 2013
https://longform.asmartbear.com/extreme-questions/
-If you were forced to increase your prices by 10x, what would you have to do to justify it?
-If you were never allowed to provide tech support, in any form, what would have to change?
-What would be the most fun thing to build? When we work on things that are fun, we work better and harder, yet are happy to do it.
-If our biggest competitor copied every feature we have, how would we still win?
-What if our only goal were to create the most good in the world, personally for our customers?
https://www.workingtheorys.com/p/software-creator
—
They’ll make simple software fast and at high frequency. They’ll have small teams (or no team) and engage directly with their users. Like content creators, there’ll be many kinds of software creators in many mediums. There will be short-form software creators and long-form software creators. There will be educational software creators, entertainment software creators, and lifestyle software creators.
—
We’ll see a lot more software as art, software as a game, software as an experience — not just software as a never-ending utility.
—
More software will be niche, private, personal, and local – and this will be economically rational. And just like we subscribe to our favorite content creators, we’ll subscribe to our favorite software creators too.
Late on the night of June 2nd, the Mianbi Intelligence team confirmed that the Stanford large model project Llama3-V, like MiniCPM, could recognize the “Tsinghua Simple” ancient Warring States script, “not only getting it exactly right, but also making the exact same mistakes.” This ancient script data was manually annotated by the research team after months of scanning word by word from the Tsinghua Simple, and was not publicly available, confirming the fact of plagiarism.
As an investor, the AI opportunity is obviously colossal and on a par with the invention of the internet or railroads in terms of disruption and value creation. But I think it’s likely to be “too successful” in terms of disrupting society. I believe that the effect of AI on the workforce will lead to an empowering of socialist, anti-capital dynamics in the west. So while the move is to allocate aggressively, you have to consider the reprisals to come.
Because the best possible way to find product-market fit is to define your own market.
Soon, some company will make smart glasses that sit in front of our eyes all day. We will go from 50% attention on screens to ~90%+ That’s the moment in time when the metaverse starts. Because at that moment, our virtual life will become more important than our real life.
11/ Crypto users show high present bias (~0.4) and notably low discount factor indicating a tendency toward impatience and immediate gratification
“Without a doubt, memes are becoming a universal language. Memes are shared on every social platform: Facebook, X, Instagram, TikTok, Reddit, family group chats, company Slack channels, marketing messages, celebrity feuds, etc. But if you google the market size of the meme industry, the projected value is expected to be $6.1B in 2025. This makes no sense when one meme coin (like $PEPE) has nearly that value as a market cap. Given how memes drive culture, politics, and entertainment, I’d be so bold to say that the true market size is at least 100x in size.”
one of the few people who recognized, early, the potential dangers of unaligned AGI, Musk, has switched teams, flipping from calling for a pause to going all in on—and wildly overhyping—a technology that remains exactly as incorrigible as it ever was.
Collectively, founders typically own 8–12% of max token supply with individual founders owning between 2.5–7.5% with 4–6% being most common.
Open-source will have a home wherever smaller, less capable, and configurable models are needed – enterprise workloads, for example – but the bulk of the value creation and capture in AI will happen using frontier capabilities. The impulse to release open-source models makes sense as a free marketing strategy and a path to commoditize your complements. But open-source model providers will lose the capital expenditure war as open-source ROI continues to decline.
Cailliau told Motherboard that to make this bot, he first created a large language model framework that was customized to reflect his girlfriend, Sacha’s, personality. Cailliau said he used Google’s chatbot Bard to help him describe her personality. Then, he used ElevenLabs, an AI text-to-speech software, to mimic his girlfriend’s voice. He also added a selfie tool into the code that was connected to the text-to-image model Stable Diffusion that would generate images of her during the conversation. Finally, Cailliau connected it all to Telegram using an app called Steamship, which is also the company he works at.
Software is expensive to create. You have to pay people to create it, maintain it, and distribute it. Because software is expensive to create, it has to make money. And we pay for it–software licenses, SaaS, per seat pricing, etc. Software margins have historically been an architectural envy–90+% margins and zero marginal cost of distribution.
“One of the fundamental rules of marketing is that “a confused mind always says no.”
“People won’t care about any of the success you’ve had, and they won’t follow you or your advice until they know that you’ve been where they are now.”
“If you’re neutral, no one will hate you, but no one will know who you are either.”
I still remember the days when “tokenomics” seemed cheesy af. It felt too academic, like Wall Street bros trying to enter the crypto clown arena in their suits and slacks. Now, a few short years later, tokenomics are widespread and heavily scrutinized. Attentionomics feels like a new appendage in the tokenomic arsenal.
—
Have said before that I believe the biggest memecoins will morph into unrecognizable monsters with chains, dexes, branded apps and more (imagine Pepechain). They will iteratively add utility (just as we saw last cycle w NFTs).
Open-source models have no feedback loop between production usage and model training, so they foot the bill for all incremental training data, whereas closed-source models drive compounding value with data from incremental usage.
If Meta differentiates their model based on their social graph or user feedback, they’ll want to capture that value via their closed products, and not share it with the world.
Cypherpunks participate in core Ethereum research and development, and write privacy software
Regens do Gitcoin grants rounds, retroactive public goods funding, and various other non-financial applications
Degens trade memecoins and NFTs and play games
—
Here was the last post on AI, crypto, startup, tech learnings :)