ImaginaryCoinPotential

[ImaginaryCoinPotential @IC_potential]

Prediction. 1 year from now Caffeine will not run on the ICP blockchain.

It will run on a cloud engine version of the ICP replica owned by Caffeine, running on AWS.

Burning zero cycles or ICP.

[dom | icp @dominic_w]

@zucaro_mar83047 The ICP replica/node has been reengineered to run on AWS instances in addition to specialized ICP hardware, so node providers will be able to create nodes on AWS.

AWS was incredibly helpful to DFINITY engineering teams and provided special functionality.

No deal needed.

[ImaginaryCoinPotential @IC_potential]

@yhabibf What % of NNS users use the new UI vs. the old UI?

[ImaginaryCoinPotential @IC_potential]

Well said

[Bubblegum Bitch @RetroDiscoFreak]

@dominic_w Wtf is Motoko

[ImaginaryCoinPotential @IC_potential]

@Missing_U_4Ever @OWHistory Knowing Dom it will be filed in the coming days or weeks if not already

[ImaginaryCoinPotential @IC_potential]

@HeyItsMeTheDev @_bluexica Point your LLM at https://t.co/LzeJ4FM0AN and deploy your first app there.

You’d be surprised with the types and complexity of apps that you can host completely on the blockchain.

[(((Michael Steuer))) - michael.cspr @mssteuer]

This story is getting more interesting, after yesterday's revelation that Near privacy is essentially built as a centralized service, not protocol native privacy.
Coincidentally, Casper also had a lead engineer "hired away" by Near in 2021

[ImaginaryCoinPotential @IC_potential]

Can’t look away 🫣 …too spicy 🌶️

[dom | icp @dominic_w]

Dude, you sum up your kind. Rather than answer the points directly, you respond like the filthy money-gribber you are with ad hominem attacks on me, DFINITY and the ICP community. You even use how the ICP price, manipulated on FTX, dropped at launch in 2021, aligning yourself with, and promoting, the work of serious criminals who are in jail. Take a look in the mirror. You're sad.

[ImaginaryCoinPotential @IC_potential]

@SnassyIcp @dominic_w The follow-through

[dom | icp @dominic_w]

https://t.co/3SY3hbel7T

[ImaginaryCoinPotential @IC_potential]

@sagewilyam @OWHistory He’s pretty toxic to work with. Surrounds himself with “yes” people and gets rid of people who push back/disagree.

Had the money to hire and keep really great talent initially but DFINITY’s runway has dwindled - it’s not what it once was.

[ImaginaryCoinPotential @IC_potential]

@sagewilyam @OWHistory Many people got f’d by dom and enjoy trolling. It’s cathartic, and for those that still see promise in the tech/original vision but who are tired of Dom…we get a kick out of it.

[ImaginaryCoinPotential @IC_potential]

This is a fair assessment of Bon’s critique.

The ICP dashboard should definitely stop counting queries as “transactions”, or “Ethereum equivalent transactions”.

As for the rest of it…other blockchains can’t asynchronously compute like ICP - one can make a case to count those.

[Sage Williams @sagewilyam]

I ran Bons thread through Gemini Argon on High. For those interested here is what Gemini says:

Justin Bons gets the core numbers right, even if his framing is aggressive. His breakdown exposes a real gap between how DFINITY markets network activity and how the rest of the blockchain industry measures transactions. At the same time, parts of his critique overlook why the Internet Computer Protocol (ICP) records internal activity differently than monolithic chains like Ethereum or Solana.
1. The 6,000 TPS Headline Figure and Read Queries
The Claim: DFINITY reports over 6,000 transactions per second (TPS) on its official dashboard, but roughly 4,000 of those are read only queries.
The Reality: This claim is accurate. The headline counter on the ICP dashboard (which sits around 5,890 to 6,070 TX/s as of October 2026) combines state changing update calls with read only query calls.
On ICP, a query call reads data from a single node without going through consensus or modifying the ledger state. Every other major blockchain handles read requests through off chain RPC nodes (such as an eth_call on Ethereum) and never counts them as transactions. Bundling read queries into a headline transaction metric inflates the number and makes direct comparisons with other networks misleading.
2. The 2,000 Update Calls vs. 0.4 User TPS
The Claim: The remaining 2,000 TPS consists of internal application steps like contract to contract calls, replies, timers, and heartbeats. Once those are removed, actual user transactions sit between 0.4 and 0.7 TPS.
The Reality: The math is accurate, and it highlights a fundamental architectural difference in how blockchains define a transaction.
On Ethereum or Solana, a user transaction executes synchronously inside a single atomic block. If a user calls Contract A, and Contract A calls Contract B and receives a response, that entire chain of events counts as one transaction.
ICP uses an asynchronous actor model where smart contracts (canisters) pass messages to each other across independent subnets. Because these steps happen asynchronously across different consensus rounds, ICP tracks each execution step as a separate update message:
* Ingress messages: External transactions signed and submitted directly by users. This is the direct equivalent of a standard transaction on Ethereum or Solana. Filtering strictly for ingress messages and token ledger transfers brings ICP down to roughly 0.4 to 0.7 TPS.
* Canister to canister calls: When one canister sends an instruction to another canister, the network processes it as a distinct consensus message.
* Replies and callbacks: When a called canister finishes its work and sends data back to the first canister, that reply is also processed as a separate update message. That is why replies alone account for more than half of all update traffic on the network.
* Timers and heartbeats: ICP canisters can execute scheduled code automatically on every block (heartbeats) or at set intervals (timers) without an external user triggering them.
Bons is right that comparing 2,000 internal execution messages on ICP to 2,000 user signed transactions on Solana is an apples to oranges comparison. Calling those internal messages completely fake, however, misses a key technical detail. Every inter canister call and timer runs through subnet consensus, mutates state, and pays for compute cycles. On Ethereum, smart contracts cannot wake themselves up, so protocols rely on external keeper bots (such as Chainlink Automation) to submit transactions that poke the contract. Those bot pokes count as user TPS on Ethereum, whereas ICP handles them natively at the protocol level through timers.
3. The 2022 Developer Forum Incident
The Claim: In 2022, a developer running less than one database upload per second generated 1,300 TPS on a subnet because each upload triggered 3,000 internal messages.

[ImaginaryCoinPotential @IC_potential]

You can onramp to ICP/crypto without going through a CEX now

[dom | icp @dominic_w]

@bitetoearn @JoshHQ @BlockchainPill The cycles you are buying are created from ICP (i.e. when you pay using your cc, behind the scenes, you are buying ICP, that's being transformed to cycles). This is a convenience layer over the decentralized network that removes token processing friction for mainstream users.

[ImaginaryCoinPotential @IC_potential]

@dominic_w @bitetoearn @JoshHQ @BlockchainPill So you guys built a crypto onramp?

Now I can convert dollars to cycles, then trade cycles back to icp on @ICPSwap through the cycle ledger DFINITY built, then go from ICP -> ckBTC -> BTC

[ImaginaryCoinPotential @IC_potential]

@sagewilyam @Oscar_ivs @dominic_w Just use the codex portion of ChatGPT. You should be able to do everything through the desktop app

[ImaginaryCoinPotential @IC_potential]

@ajki76 @plsak @OWHistory Those are just API definitions.

The services that power those APIs are not open sourced, so we have no idea how the metrics are derived from the data. We trust that DFINITY is honest and that their current interpretation of each metric represents is reasonable and correct.

[ImaginaryCoinPotential @IC_potential]

@SnassyIcp @devXritesh You’d store a map of the valid jwts in a
cache and delete on logout so doesn’t exist - check fails validation

[ImaginaryCoinPotential @IC_potential]

@plsak @ajki76 @OWHistory If only the code for the dashboard were open sourced we’d know the answer to this

[ImaginaryCoinPotential @IC_potential]

.icp TLD

Hopefully it gets approved!

[ImaginaryCoinPotential @IC_potential]

@sagewilyam @JerryBanfieldIC It’s honestly nuts to think about how big of a whiff Caffeine was. Instead of just focusing on being the blockchain with the best tools and easiest agent integration, @dominic_w tried to build the AI harness and outcompete big tech, all while relying on Claude under the hood.

[ImaginaryCoinPotential @IC_potential]

@sagewilyam @JerryBanfieldIC Completely agree.

Blockchain/crypto still works as a frictionless payment layer that 2 parties’ agents can plug into to coordinate a transaction, but Muse has proven out web2 agent PMF - people are more than willing to hand their credit card to an agent and let it go to town.

[ImaginaryCoinPotential @IC_potential]

@mosesibb These types of projects have done a lot of harm to ICP retail investors.