_weidai
0 Suivis
3 Abonnés
3 J’aime
0 Partagé(s)
Publications
MVD -- minimum viable decentralization
Are there any reasons why an app operated by a company is better than a decentralized app?
Also suppose that latency <500ms doesn't matter for this app and other features like scaling, cost, and privacy are all at parity
Assume this app need to take in deposits (of some form)
Also suppose that latency <500ms doesn't matter for this app and other features like scaling, cost, and privacy are all at parity
Assume this app need to take in deposits (of some form)
Crypto Needs Flat Base Fees and Dynamic Scaling
Imagine if Google charged you 10x normal price for every search, LLM inference, or youtube video because they run out of servers and hit some internal capacity.
That's how EIP1559 and most blockspace pricing mechanisms work--in times of congestion, instead of increasing capacity (like sane business do), chains raise prices instead.
Recall there are two types of fees:
- Inclusion/base fee: fee required for a transaction to be included.
- Priority fee: fees for getting in front of the queue to access contentious state (e.g. hot onchain markets)
If onchain apps is to seriously compete with non-crypto apps, we need flat inclusion/base fee for every chain for any transaction demand.
Q: What's the problem if chains adopt constant base fees?
A: Capacity. A constant base fee means that there is no way to discriminate against incoming transactions once chains hit a limit.
Solution? Dynamically scaling!
Simply process more transactions as they come through, without a global limit.
For too long now we have relied on crypto-economists to solve the fee vs. chain capacity problem. It is now the time to let computer scientists and engineers do the work--we need to build chains dynamically scale.
Ask yourself what is the better product decision for endusers:
- Increase fees at times of congestion
- Add compute capacity to the system to accommodate demand
The answer is self-evident. We've been building the wrong product for more than a decade!
How do we do this? We need to build better system at every layer:
- Authenticated data structures that can horizontally scale without global bottlenecks
- Execution models & VMs that allow infinite parallelism
- Distributed consensus that allow blocks to be dynamically sized
We need flat fees for data (blobs) and execution (gas/CU).
Dynamic scaling is the future for crypto.
Imagine if Google charged you 10x normal price for every search, LLM inference, or youtube video because they run out of servers and hit some internal capacity.
That's how EIP1559 and most blockspace pricing mechanisms work--in times of congestion, instead of increasing capacity (like sane business do), chains raise prices instead.
Recall there are two types of fees:
- Inclusion/base fee: fee required for a transaction to be included.
- Priority fee: fees for getting in front of the queue to access contentious state (e.g. hot onchain markets)
If onchain apps is to seriously compete with non-crypto apps, we need flat inclusion/base fee for every chain for any transaction demand.
Q: What's the problem if chains adopt constant base fees?
A: Capacity. A constant base fee means that there is no way to discriminate against incoming transactions once chains hit a limit.
Solution? Dynamically scaling!
Simply process more transactions as they come through, without a global limit.
For too long now we have relied on crypto-economists to solve the fee vs. chain capacity problem. It is now the time to let computer scientists and engineers do the work--we need to build chains dynamically scale.
Ask yourself what is the better product decision for endusers:
- Increase fees at times of congestion
- Add compute capacity to the system to accommodate demand
The answer is self-evident. We've been building the wrong product for more than a decade!
How do we do this? We need to build better system at every layer:
- Authenticated data structures that can horizontally scale without global bottlenecks
- Execution models & VMs that allow infinite parallelism
- Distributed consensus that allow blocks to be dynamically sized
We need flat fees for data (blobs) and execution (gas/CU).
Dynamic scaling is the future for crypto.
Tech defines the potential of onchain apps.
Building onchain apps on shitty tech/infra is like hosting YouTube on dialup modems.
Without good tech or infra, onchain apps won't scale to billions regardless of how good your distribution is.
Tech tradeoffs between latency, throughput, security, privacy, are not at all commoditized yet.
No, you can't vibe code a competitive, decentralized, and self-custodial CLOB, casino, or social app.
Building onchain apps on shitty tech/infra is like hosting YouTube on dialup modems.
Without good tech or infra, onchain apps won't scale to billions regardless of how good your distribution is.
Tech tradeoffs between latency, throughput, security, privacy, are not at all commoditized yet.
No, you can't vibe code a competitive, decentralized, and self-custodial CLOB, casino, or social app.
Tech defines the potential of onchain apps.
Building onchain apps on shitty tech is like hosting YouTube on dialup modems.
Without good tech, onchain apps won't scale to billions regardless of how good your distribution is.
Tech tradeoffs between latency, throughput, security, privacy, are not at all commoditized yet.
No, you can't vibe code a competitive, decentralized, and self-custodial CLOB, casino, or social app.
Building onchain apps on shitty tech is like hosting YouTube on dialup modems.
Without good tech, onchain apps won't scale to billions regardless of how good your distribution is.
Tech tradeoffs between latency, throughput, security, privacy, are not at all commoditized yet.
No, you can't vibe code a competitive, decentralized, and self-custodial CLOB, casino, or social app.
Not going to Berlin blockchain week.
But will be at Edge Esmeralda next week (15-20th).
HMU if you are working on interesting problems.
But will be at Edge Esmeralda next week (15-20th).
HMU if you are working on interesting problems.
Recent improvements in crypto infra are getting incremental: we have steady improvements on latency, throughput, faster zk proofs, etc.
What are some exciting open challenges that, if solved, enable net new apps?
(Privacy is one for sure but curious of others.)
What are some exciting open challenges that, if solved, enable net new apps?
(Privacy is one for sure but curious of others.)
1/ "Zero trust" is a term that is growing fast in popularity, with mindshare growing faster (per Google trends) than "decentralization" and "self custody", two main pillars of crypto.
What is Zero Trust?! 🧵
What is Zero Trust?! 🧵
Why crypto need to refocus on capability, over blind emphasis on decentralization or verifiability.
All technological progress are defined by the capabilities unlocked.
For crypto:
- Censorship-resistant money (and store of value) was the first capability unlock.
- Permissionless asset creation and finance was the second capability unlock.
Notice the theme: censorship resistance and permissionless-ness, which are security features enabled by decentralization, were critical to these first two unlocks--
- Bitcoin would not be what it is if not for its self-custodial property and censorship resistance.
- Ethereum would not have been what it is without ICO and defi summer, which were enabled by permissionless smart contracts.
Over the last few years, crypto got sidetracked and started to value decentralization, security, and verifiability without specific capability unlocks. Examples: decentralization theater (early forms of onchain AI), adding verifiability without specific purpose ("zk-everything").
I believe our the industry is finally exiting this sidetrack.
Example: If an app requires real-time censorship resistance, then decentralize. No need for real-time censorship resistance? Let's use a centralized sequencer but offer self-custodial guarantee via validity proofs.
Refocus on capability; make crypto great again.
(Image credit ChatGPT 4o)
All technological progress are defined by the capabilities unlocked.
For crypto:
- Censorship-resistant money (and store of value) was the first capability unlock.
- Permissionless asset creation and finance was the second capability unlock.
Notice the theme: censorship resistance and permissionless-ness, which are security features enabled by decentralization, were critical to these first two unlocks--
- Bitcoin would not be what it is if not for its self-custodial property and censorship resistance.
- Ethereum would not have been what it is without ICO and defi summer, which were enabled by permissionless smart contracts.
Over the last few years, crypto got sidetracked and started to value decentralization, security, and verifiability without specific capability unlocks. Examples: decentralization theater (early forms of onchain AI), adding verifiability without specific purpose ("zk-everything").
I believe our the industry is finally exiting this sidetrack.
Example: If an app requires real-time censorship resistance, then decentralize. No need for real-time censorship resistance? Let's use a centralized sequencer but offer self-custodial guarantee via validity proofs.
Refocus on capability; make crypto great again.
(Image credit ChatGPT 4o)
Dee Hock on the founding of Visa in his work "Chaordic Organization".
Visa was perhaps the first successful case of a permissionless platform with decentralized governance and ownership.
Many lessons to be learned here half a century later for crypto networks.
Visa was perhaps the first successful case of a permissionless platform with decentralized governance and ownership.
Many lessons to be learned here half a century later for crypto networks.
prediction: the word "dapp" will die
just apps or onchain apps
just apps or onchain apps
A straddle (call & put option or a long & short perp) is like quantum teleportation for money.
Value magically goes from one position to the other as price of the underlying moves.
Don't believe PnL charts of a single account.
Value magically goes from one position to the other as price of the underlying moves.
Don't believe PnL charts of a single account.
Reject privacy maximalism; embrace privacy pragmatism.
1/ The Onchain Privacy Trilemma 🧵👇
Privacy is the last hurdle to mass-adoption of onchain finance. Can we simply add self-sovereign privacy to today's onchain finance and expect everything to work better?
The answer is no. We must make hard decisions on trade-offs between the following three desirable properties:
1. Maximally useful: no transaction limits, supports private payments and anonymous DeFi
2. Self-sovereign privacy: contents of private transactions cannot be revealed without consent from involved individuals
3. Threat resistant: adversaries cannot use it to hack & launder funds
Choose any 2, but never 3
1/ The Onchain Privacy Trilemma 🧵👇
Privacy is the last hurdle to mass-adoption of onchain finance. Can we simply add self-sovereign privacy to today's onchain finance and expect everything to work better?
The answer is no. We must make hard decisions on trade-offs between the following three desirable properties:
1. Maximally useful: no transaction limits, supports private payments and anonymous DeFi
2. Self-sovereign privacy: contents of private transactions cannot be revealed without consent from involved individuals
3. Threat resistant: adversaries cannot use it to hack & launder funds
Choose any 2, but never 3
How many onchain crypto apps have hit PMF & hyper-growth phases? A: a few.
Out of those, how many sustained the demand without hiccups? A: practically none.
Solving blockchain scaling is about empowering apps to reach PMF & hyper growth.
Out of those, how many sustained the demand without hiccups? A: practically none.
Solving blockchain scaling is about empowering apps to reach PMF & hyper growth.
If an L1 enforces censorship after a large hack, then why not go all the way to implement in-protocol mechanisms to enforce social consensus so that any consensus hack can be reverted?
In the end-state a chain is either a neutral base layer or a social consensus engine.
In the end-state a chain is either a neutral base layer or a social consensus engine.
Blob gas price has dropped back to 0 after Pectra.
Uber doesn't charge $0 when there are more drivers than riders. Why would Eth charge $0 for a valuable resource?
What if EIP1559 is actually terrible for non-congested markets as it was designed for congested ones?
Uber doesn't charge $0 when there are more drivers than riders. Why would Eth charge $0 for a valuable resource?
What if EIP1559 is actually terrible for non-congested markets as it was designed for congested ones?
Theory crafting: SoV assets need *stability* of cashflow/REV, rather than the lack of it.
===========
If BTC (or Gold) started to generate cashflow, it won't stop being a SoV asset, at least right away.
But, if this cashflow ever decrease, the valuation would drop in response.
Plus, markets may over-index on cashflow falling: a SoV asset with falling cashflow is less appealing than another SoV whose cashflow is not falling (could be due to it being zero).
Compound that with the fact that SoV relies on network effects, which means that relative marketshare movements could get amplified (a winning SoV can win harder).
===========
In upshot, the downside of cashflow/REV for a SOV asset is that it makes the asset less appealing when cashflow falls.
Therefore, what's really important for SoV assets is the *stability* of cashflow/REV, rather than the lack of it.
(All of this is mostly empty speculation from first principles and not backed by any real data btw. So take it with a grain of salt.)
===========
If BTC (or Gold) started to generate cashflow, it won't stop being a SoV asset, at least right away.
But, if this cashflow ever decrease, the valuation would drop in response.
Plus, markets may over-index on cashflow falling: a SoV asset with falling cashflow is less appealing than another SoV whose cashflow is not falling (could be due to it being zero).
Compound that with the fact that SoV relies on network effects, which means that relative marketshare movements could get amplified (a winning SoV can win harder).
===========
In upshot, the downside of cashflow/REV for a SOV asset is that it makes the asset less appealing when cashflow falls.
Therefore, what's really important for SoV assets is the *stability* of cashflow/REV, rather than the lack of it.
(All of this is mostly empty speculation from first principles and not backed by any real data btw. So take it with a grain of salt.)
Theory crafting: SoV assets require *stability* of their cashflow/REV, rather than the lack of it.
===========
If BTC (or Gold) started to generate cashflow, it won't stop being a SoV asset, at least right away.
But, if this cashflow ever decrease, the valuation would drop in response.
Plus, markets may over-index on cashflow falling: a SoV asset with falling cashflow is less appealing than another SoV whose cashflow is not falling (could be due to it being zero).
Compounds that with the fact that SoV relies on network effects, which means that relative marketshare movements could get amplified (a winning SoV can win harder).
===========
In upshot, the downside of cashflow/REV for a SOV asset is that it makes the asset less appealing when cashflow falls.
Therefore, what's really important for SoV assets is the *stability* of cashflow/REV, rather than the lack of it.
(All of this is mostly empty speculation from first principles and not backed by any real data btw. So take it with a grain of salt.)
===========
If BTC (or Gold) started to generate cashflow, it won't stop being a SoV asset, at least right away.
But, if this cashflow ever decrease, the valuation would drop in response.
Plus, markets may over-index on cashflow falling: a SoV asset with falling cashflow is less appealing than another SoV whose cashflow is not falling (could be due to it being zero).
Compounds that with the fact that SoV relies on network effects, which means that relative marketshare movements could get amplified (a winning SoV can win harder).
===========
In upshot, the downside of cashflow/REV for a SOV asset is that it makes the asset less appealing when cashflow falls.
Therefore, what's really important for SoV assets is the *stability* of cashflow/REV, rather than the lack of it.
(All of this is mostly empty speculation from first principles and not backed by any real data btw. So take it with a grain of salt.)
Theory crafting: SoV assets requires *stability* of their cashflow/REV, rather than the lack of it.
===========
If BTC (or Gold) started to generate cashflow, it won't stop being a SoV asset, at least right away.
But, if this cashflow ever decrease, the valuation would drop in response.
Plus, markets may over-index on cashflow falling: a SoV asset with falling cashflow is less appealing than another SoV whose cashflow is not falling (could be due to it being zero).
Compounds that with the fact that SoV relies on network effects, which means that relative marketshare movements could get amplified (a winning SoV can win harder).
===========
In upshot, the downside of cashflow/REV for a SOV asset is that it makes the asset less appealing when cashflow falls.
Therefore, what's really important for SoV assets is the *stability* of cashflow/REV, rather than the lack of it.
(All of this is mostly empty speculation from first principles and not backed by any real data btw. So take it with a grain of salt.)
===========
If BTC (or Gold) started to generate cashflow, it won't stop being a SoV asset, at least right away.
But, if this cashflow ever decrease, the valuation would drop in response.
Plus, markets may over-index on cashflow falling: a SoV asset with falling cashflow is less appealing than another SoV whose cashflow is not falling (could be due to it being zero).
Compounds that with the fact that SoV relies on network effects, which means that relative marketshare movements could get amplified (a winning SoV can win harder).
===========
In upshot, the downside of cashflow/REV for a SOV asset is that it makes the asset less appealing when cashflow falls.
Therefore, what's really important for SoV assets is the *stability* of cashflow/REV, rather than the lack of it.
(All of this is mostly empty speculation from first principles and not backed by any real data btw. So take it with a grain of salt.)
Connectez-vous pour découvrir d’autres contenus