PROMPTS ARE TINY PROGRAMS
We’re now about 18 months into the AI revolution. One thing that was uncertain in late 2022 was whether prompt engineering would be around to stay, or whether better AI would quickly obviate it.
I now think it’s around to stay and I have an explanation that makes sense to me, at least: prompt engineering is just a subset of software engineering.
That is, prompts are tiny programs written in natural language. But the API isn’t specified and varies between models. So guessing the right “function calls” with clever use of vocabulary is a huge part of the game. On the other hand, even if you don’t guess *exactly* the right words to use, the model will often do what you mean.
This is different from how we normally think of an API, which is both more legible and more fragile. The exact words to make an API do what you want are written down, but if you don’t say those exact words it won’t do what you want.
Even given this difference, the concept of prompts as tiny programs using hidden APIs helps explain the bizarre magic associated with specific phrases. I’m reminded of Quake3’s fast inverse square root[1,2], which has a famously obscure incantation in C that just so happened to deliver a 4X speedup.
More code now looks like that, and it makes sense. C is how you talk to machines and English is how you talk to humans. So, just like you write part of a large application in C for performance, you’ll also write part of it in English for dealing with unstructured data.
You can go further with this analogy. Once you think of prompts as code, you can probably generate model-aware syntax highlighters for favored keywords. You can maybe automatically generate API-like docs from a model for the most common use cases.
And you can think of every new model you add to your codebase as roughly analogous to adding a new programming language — because just as it takes time for someone to ramp up on the idioms of Rust, they’ll need to play around with the latest Mistral to get the hang of how to talk to it.
Anyway — this is all probably obvious to folks spending 100% of their time in the field, and is similar to some of the things @karpathy has posted about, but at least for me it was a useful articulation of why prompts are around to stay: prompts are tiny programs.
[1]: https://t.co/3Ryz3N69hY
[2]: https://t.co/qK2C7IU2g1
Reading a tweet is a bit like downloading an (attacker-controlled) executable that you instantly run on your brain. Each one elicits emotions, suggests knowledge, nudges world-view.
In the future it might feel surprising that we allowed direct, untrusted information to brain.
@ShibuyaProd This is where I learned my 2nd lesson.
→ "users" are cool but worthless if no one is paying.
If people are paying you, it's easier to identify a product market fit and scale from there.
Spending 2 years just building an app doesn't make sense.
David's Take 🗣️
Is this a different type of Bull Market?
If it is, we're missing crypto's homegrown catalyst 👇
---------------
Apparently, it’s a bull market
At least, that’s what everyone is saying
And this shouldn’t be discounted!
If a critical mass of people is bullish, that sentiment becomes contagious, and the time to bid is now!
But there’s something different about this bullish sentiment from previous cycles, and it’s worth identifying and parsing out
People are bullish because of the Spot ETFs that are imminent
The Bitcoin Spot ETF approval could happen anytime, and the Spot ETH ETF isn’t far behind
This is a moment this industry has been waiting for over a decade -- the Winkelvi first filed their Bitcoin ETF in 2013!
The size of the pipe that a Spot ETF builds between BTC and ETH should not be underestimated
A TON of money can flow through these pipes
It’s a massively bullish catalyst, and there’s no other way to interpret that
If these Spot ETF approvals triggered a bull market, it would be the first time crypto experienced an externally caused bull market
What do I mean by this?
- The 2013 bull market was known as the Fork-and-Fair-Launch era, where people realized that Bitcoin was a primitive that could be copied. 10,000 PoW chains blossomed, and a few of them are still alive to this day
- The 2017 bull market was the ICO mania, where people realized that you could just issue a token on Ethereum, and you wouldn’t need to have your asset be encumbered by an entire blockchain. Plus, smart contracts!
- The 2021 bull market was the NFT mania and the Alt-L1 mania, in which we figured out that our tokens didn’t have to necessarily be just financial assets, but culture, art, and community were able to realize value in DeFi. Also, we needed more blockspace, and the desire to produce an “Ethereum-killer” attracted a ton of VC funding and energy
- The 2024 bull market currently seems poised to be the spot ETF approval bull market?
Because external capital is capable of buying BTC and ETH?
Do you notice something different here?
This bull market is not on our terms
This is an exogenous bull market
It’s being caused by external forces beyond the crypto industry
There is no internal catalyst for excitement, interest, and blockspace demand
In all previous crypto bull markets, we’ve found some new primitives that unlocked new use cases, new applications, and new demand
We excited consumers and retail with new powers and drew people in with an expanded vision of what crypto could be
In this bull market, TradFi can purchase our two main blue-chip assets in their brokerages
These are different bull market catalysts
Right now, bullish sentiment has gone viral in the industry, and down-market cap assets have experienced some insane price actions, notably THORChain and Solana, two assets that also had huge pumps last bull market
Industry veterans who stuck around all bear market are putting their chips on the table, placing their bets on what a retail speculative frenzy might look like, and are using the memory of 2021 to inform their investment decisions
But what if this bull market isn’t like 2021?
Traders and speculators are using historical activity and price action to inform what might attract users in 2024
Still, all the signals we have right now point to a different kind of bull market than one we previously had
A Tale of Two Markets
With the incoming Spot BTC and ETH ETFs, there will be a big fat line drawn between the crypto-assets that have Spot ETF exposure and those that do not
The reason why the Spot ETFs are bullish is due to the massive amounts of exposure that create the trillions of dollars of wealth held in traditional brokerages and the rest of TradFi
Only BTC and ETH have this exposure!
People placing their chips below this line are speculating on a bull market that matches the historical bull markets that crypto has experienced. But, without an endogenous catalyst of internal excitement, new interest, and new capital, why would the bull market proceed past this big line between BTC, ETH, and the rest of the market?
We need an Endogenous Catalyst
Bankless Nation, like you, I want a bull market
But if we’re going to have the “bull market” that we all imagine, we need a reason for new entrants to come into crypto directly via private keys and start playing in the arena below BTC and ETH
Right now, we don’t have any new reason for people to enter crypto in a crypto-native way
We haven’t unlocked any new primitive
Retail doesn’t care that Polygon or Solana has cheap blockspace and improved execution environments
We haven’t built any new apps!
Now, increasing asset prices is marketing in itself, and the Casino re-opening and people making money for the first time in 2 years is an incentive to re-enter the arena
But I fear that that is short-lived if there’s not some new app or primitive that people can play with, as there was in 2013, 2017, and 2021
To have the bull market we all want, we need an endogenous catalyst
DAOs of the future will be builders not fund allocators
1. They won’t start with a treasury or a token but with like minded peeps who share vibes, excitement and skills
2. They’ll be small, highly aligned and efficient
3. They’ll build in public and be community first around a more organic community
4. They’ll decentralize slowly using soulbound NFTs with peeps who bring real value to the project
5. Supported by LLMs these DAOs will move fast and build things
6. Also emergence of a new class of DAOs, task based that are micro and ephemeral. Task done, DAO done
The future DAOs will be builders
This is a superb reflection on the state of AI vs Human Writing. The upshot? A nuanced picture. Thanks Amy Suto for the candid and thoughtful perspective.
#writing#ai#terminator https://t.co/kwuv4ZSP0B
Fiat-backed crypto is fast becoming the leading use-case for blockchain technology. All the strengths of fiat, with none of its weaknesses - the "Blade" of international finance.
<insert Wesley Snipes pouting menacingly with a sa…https://t.co/Y7cKedGSeQ https://t.co/LF0213zVSc
Web3, Fintech, and Wealth Management all need each other.
🎯 Wealth needs a way to organic growth and lower costs for a low interest rate age.
🎯 Web3 needs a path to real-world relevance, and a user base who will adopt it.
🎯 Fintech needs a big prob…https://t.co/MQbB212lK0
Recommunitizing the globe could be messy, but ultimately better for everyone's mental health. And it could turn out to be crypto's main raison d'etre.
https://t.co/Fxl6dXm77B