Tom Howard

Post-Sovereign Chief Biomaximizer.

Foundation for ending Foundations.

If I detect AI responses to my post I will kill myself racistly.

[Tom Howard @_TomHoward]

The Zcash resurrection has been underway for a few years. People are finally discovering it today.

Follow the early movers.

https://t.co/dpb7wbu101

[Tom Howard @_TomHoward]

@arkham Your source that Zcash was down is your own incomplet dataset which is MISSING DATA.

The problem is actually that Arkham went down for hours.

Complete data source: https://t.co/xXRuHHJQEE

[Tom Howard @_TomHoward]

Probably not a thing

[Alex Svanevik 🐧 @ASvanevik]

if only there was a version of BTC

that's private

with no Saylor overhang

and priced 100x lower

[Tom Howard @_TomHoward]

@bonsaiquant @tulipking @notthreadguy And thats why you aren't a north korean hacker for a living

[Dev 🧪 @zkDragon]

So what you can formally verify is that "this code meets all properties of this specification".

Now a lot of the code we care about has very precise specifications. "This ZK-circuit is satisfied only if {these variables}, follow the exact scalar multiplication relationship".

Now this often sounds odd on first pass.
(1) How do you express the specification to the formal prover/verification?
(2) Do you end up just ... rewriting the code to prove it, so your not actually proving your real code?
(3) Did your spec account for everything.

(1) We can make the spec something like this:
```
Given:
leaf
authentication path

Output:
root

such that root = MerkleTreeHash(leaf, path)
```

And then we prove that "this R1CS circuit, if satisfied, implies that the prover's "leaf,path,root" satisfy `root = MerkleTreeHash(leaf, path)`.

(2) We must compile our real circuit to the formal verification language. We can't just rewrite the circuit in the formal verifier, that wouldn't prove anything important. This is the huge unlock of AI, where this used to be months-to-year long tasks. Now its compressed to days/weeks. I was a huge FV bear until this year, for this reason.

(3) this is always the issue. (Called out-of-model errors) Does your spec account for everything. In the case of Zcash circuits, I have high confidence we do at a spec level. If you understand whats happening, its actually quite simple in abstract terms. Just translating those hashes to efficient circuits is the hard part.

There are millions of examples of out-of-model errors. To me the most famous one is KRACK attack: https://t.co/DmHDkwMiMe

However FV has gotten incredibly clever. People have (imo shockingly!) expressed constant time, and memory safety into FV language semantics. What we need for circuits is all very simple in comparison.

[Tom Howard @_TomHoward]

I hate the way every version of Claude writes.

I don't want to talk to it but it's good at really technical code tasks.

So I have Codex tell it what to do now.

[Alex Cohen @anothercohen]

I can't explain why but I fucking hate the way Opus 4.8 writes

[Tom Howard @_TomHoward]

@tulipking @notthreadguy If it was, the shielded pool would already be drained.

[Tom Howard @_TomHoward]

@cobie @serpinxbt SpaceX and Anthropic

[Tom Howard @_TomHoward]

Shielded ZEC now accepted for Roman Storms defense fund.

The freedom to write code and encryption is at stake here!

[Roman Storm 🇺🇸 🌪️ @rstormsf]

@mert https://t.co/LOqeHvQqPx
Has unified zcash address

[Tom Howard @_TomHoward]

False.

Miners independently chose to suspend only Orchard pool txns, chain was working, other transactions always working.

Major miners got access to patch first.

Explorers each run their own RPC and did not get patch until after miners implemented (for security).

This caused explorers to go offline as they got caught up to the patch the miners were already running.

Exchanges now updating the patch that miners are running.

Nakamoto consensus is working!

Everyone decides to run new code on their own.

[Tom Howard @_TomHoward]

@TedPillows I see Ted doesn’t run his own RPC node.

[Tom Howard @_TomHoward]

@Senpai_Gideon @zkDragon Neat!

[Tom Howard @_TomHoward]

Zcash's Tachyon is not just post-quantum and post-scale, it's also post-hardened!

[Project Tachyon @TachyonZcash]

As we blogged about last month, the new shielded pool we're developing for Zcash will use formal verification, more conservative cryptography, a simple arithmetization, robust APIs, fuzzing, and extensive auditing from humans and AI.

And much more: https://t.co/H3bwCnzuLY

[Dev 🧪 @zkDragon]

Where were going:
- Every circuit formally verified
- Written in R1CS, the simplest arithmetization
- Automated fuzzing
- Over-audited by humans and an army of AI

[Tom Howard @_TomHoward]

@jswihart @zodl_app @ZcashFoundation @feministPLT @nuttycom @str4d Fixed without a hitch, well done! 🫡

[Ansem @blknoiz06]

I meme a lot but i actually do think this pair continues to go up, zcash currently trades at lower than 1/100th of bitcoin's total valuation w/ better quantum resistant properties while privacy is becoming a popular narrative

concerns w/ saylor as key man risk also materializing

should be closer to 1/10th valuation, which would be ~.1zec per btc

[Tom Howard @_TomHoward]

Whats better - one fat harness to rule them all or many purpose scoped harnesses? @garrytan?

[Thariq @trq212]

https://t.co/R6exTuF7P8

[Tom Howard @_TomHoward]

Every tech entrepreneur is learning from what the centralized AI companies are doing and building it into their own self hosted AI harness.

The margin compression on all of this is insane and the winners are the builders.

[@jason @Jason]

they are trying to kill cursor and lovable… and every startup and application — as I’ve warned

Infrastructure companies eventually try to win the platform game, then they learn and take out all their partners on the app layer

[Tom Howard @_TomHoward]

@DigitalFringed1 Listen everyone has abs it’s just hiding under the fat he’s just flipping the layers

[Tom Howard @_TomHoward]

@DrunkDialMe_ @XBToshi Interesting you used a very out of date chart to make a point that’s no longer valid