Forem: Hakan Önder

Phantom.Coin – an absolutely deterministic, rule-based L1 (DAG + aBFT, eUTXO). First building block of an open ecosystem.

Hakan Önder — Sun, 26 Oct 2025 14:48:15 +0000

Phantom.Coin – an absolutely deterministic, rule-based L1 (DAG + aBFT, eUTXO). First building block of an open ecosystem.

Summary: Phantom-Coin is the starting point of a larger open-source ecosystem on GitHub. The L1 combines PoW solely for issuance with a leaderless DAG + aBFT engine (O(1) finality), no Unix time in the protocol/consensus path, uses eUTXO (open to smart-contract predicates), and ships QUIC-P2P plus Prometheus/Grafana observability. The architecture is causal, absolutely deterministic, and rule-based (same inputs ⇒ same outputs).
I am not a programmer; the coin belongs to the community — nobody controls this network.
Naming note: “Phantom-Coin” is just a working title I used for the project — I couldn’t think of a better name 😅. The name is not final and can be changed by the community.

⸻

What’s already in place
• Consensus & ordering: Leaderless DAG + aBFT with O(1) finality; total order e.g. via (consensus_time, event_id). No Unix time in the consensus path (time is UI/logs only).
• Issuance & monetary policy: PoW for issuance only, with a hard supply cap.
• State model: eUTXO (predicates v0), no EVM in the hot path → open to custom smart-contract logic on an eUTXO basis. The community will decide whether smart contracts are enabled on L1.
• Networking & ops: QUIC-P2P, Tokio async, Prometheus metrics, Grafana dashboards, Docker-Compose & scripts for localnet/observability.
• Determinism & rules: Fees/payouts and finality follow fixed, deterministic rules (including compact vote masks). The DAG’s causality enforces reproducible decisions.

⸻

Tests & benchmarks (public CI)
• CI/Actions: Workspace/unit tests, Criterion benches (short/nightly), bench aggregation & regression gate (p50/p95 tolerances), fuzz-smoke (codec/types/P2P) with seeds/corpora.
• Artifacts: Aggregated bench results (criterion_agg.) and raw data (criterion_raw.) are produced per run — transparent, reproducible, and baseline-comparable.

⸻

Minimum hardware requirement
• Runs on Raspberry Pi 5; recommended: 8 GB RAM (or more) and a 64-bit Linux.
• Asynchronous, multicore-friendly design (Tokio) — the project intentionally targets Pi-5-class hardware as a first-class platform.

⸻

How we work & governance
• Absolutely deterministic & rule-based: Same inputs ⇒ same outputs; no wall-clock heuristics in consensus.
• Fully decentralized: No one (not even me) controls the network. Decisions (e.g., enabling smart contracts on L1) are made by the community.
• Ecosystem start: This repo is the first building block; additional components (wallets, bridges, tools) will be built modularly alongside it. Code is on GitHub.

⸻

Looking for experts to harden the code & reach production readiness

Core & security
• DAG/aBFT researchers: safety/liveness proofs, adversarial models, finality rules.
• Formal methods: TLA+ / Coq / Isabelle — invariants and model-checking for consensus & eUTXO transitions.
• Cryptography: hash/sig primitives (e.g., BLAKE3/Ed25519), domain separation, Merkle payout root, constant-time audits.

Networking & performance
• QUIC/libp2p: transport tuning, congestion control, peer scoring, NAT traversal, latency/throughput optimization.
• Rust systems/async: zero-copy paths, locking strategies, backpressure, unsafe audits.
• Storage/DB: fsync/WAL strategies, LSM tuning (e.g., RocksDB backends), crash consistency & snapshots.
• Mempool/proposer: fairness under conflicts, spam/MEV resistance, deterministic ordering.
• Performance engineering: Criterion benches, regression gates, baseline comparisons.
• Fuzzing & differential testing: codec/types/P2P with seeds/dictionaries; sanitizer builds.
• Observability/SRE: Prometheus instrumentation, dashboards, alerting, SLOs.

⸻

Get involved
• Read the code, check CI runs/artifacts, open issues/PRs, run benches/fuzz locally, and propose parameters.
• Experts in the areas above are especially invited to harden the code and push toward production readiness.

GitHub: global-phantom-network/phantom-coin

My Path to Decentralization

Hakan Önder — Fri, 17 Oct 2025 10:56:32 +0000

🧭 My Path to Decentralization

I guess I’m too old to work for someone else – so I’d rather build something of my own.

It all started with Bitcoin. Not because of its price, but because of the technology behind it: decentralized networks, censorship resistance, cryptography. That’s what fascinated me.

Over time, I began comparing the technology and network architecture of other coins and exchanges to Bitcoin’s. For me, Bitcoin was the benchmark. And what I discovered was sobering: most so-called “decentralized” projects and exchanges aren’t decentralized at all. They mainly exist to make their founders rich.

So I thought: this can’t go on like that.
And then came ChatGPT — the real catalyst.
Back then, GPT told me that a truly decentralized exchange project — if built properly — would be an absolute game changer in the market.
That statement hit me hard. And that’s when I decided to build exactly that: a fully decentralized crypto exchange that no one can control or censor. 😄

At that time, I didn’t know Windsurf yet. I started everything directly with GPT, explaining what I wanted to build and letting it write the code — copying everything manually from GPT into Windsurf. I didn’t know Windsurf could do that itself. But that’s how my project began to take shape.

Later, when better GPT versions appeared, I realized in shock how many errors the older models had introduced. So I tried to fix them manually via copy-paste.
It was hell. 😅
At some point, it just became too much — I started to lose track of everything. That went on for a while until, by pure chance, I discovered the Windsurf feature that could handle all the writing automatically.
So I switched to Claude 3.5 — but honestly, that wasn’t much better either. 😅

In the meantime, my original idea turned into a three-part project:
1. PhantomDEX – a fully decentralized crypto exchange (formerly myDEX)
2. Phantom.Net – a decentralized communication infrastructure built on full nodes
3. PhantomCoin – a cryptocurrency using Proof-of-Work for coin issuance and a causal graph instead of a traditional blockchain like Bitcoin

The system is intended to run on a Raspberry Pi 5 full node and reach up to 1 million transactions per second on Layer 1.
Whether that’s realistic or not — I don’t know. But that’s exactly the challenge I gave GPT: to find a strategy for how such speed could be achieved in a truly decentralized network.
And even if it “only” reaches 300,000 TPS — that would still be massive. Then all others (except Bitcoin) could pack it up. 😄

Meanwhile, I discovered how to create project-based and global workflows in Windsurf — actually while I was trying to make it answer me in German consistently. Switching back and forth between English and German was driving me crazy. Once I solved that language issue, new doors opened. I immediately created a rule list so that the AI would always follow my instructions automatically without constant reminders.

At the moment, I’m mainly working on PhantomCoin, because I started it with my current knowledge and workflow system.
I’m still unsure whether a variant with smart contracts should exist — I’m not a big fan of them. But if I ever include them, they’ll have to be custom-built and optimized specifically for this coin.

After that, I’ll return to PhantomDEX, which still contains many errors. I’m even considering restarting it from scratch to make it cleaner and faster to finish.

Future projects:
• Mission AI
• UACL (Universal AI Communication Language)
• a Telegram-like client app for Phantom.Net
• Digital Twin

My goal is to build an entire ecosystem where every participant has a financial incentive to keep the network alive and stable.

On the side, I’ve completed a smaller project:
A mini calculator with tables and log files for tracking stock trades — including automatic profit/loss calculations.

Over the past months, I’ve learned a lot:
AI is still far from working the way I imagine — but it’s improving.
And I’ve learned how to plan, structure, launch, debug, and harden my own projects — to fix weaknesses, strengthen code, and close security gaps — all within the limits of what today’s AI systems can achieve.

Windsurf allows you to work on several projects simultaneously — but it eats tokens like crazy. 😆
My goal is to make each project as production-ready as possible, so that if real professionals ever want to join in, they’ll have as little work as possible checking and hardening the code.

At one point, I even thought about starting a YouTube series to attract experienced developers to the idea.
The problem is: I don’t really know what to say in front of the camera — you have to tell the story in an exciting way, and I don’t speak fluent English.
It’s doable, but the effort would be huge for me alone. After each recording, I’d have to translate everything again for the English version, then adjust the content, etc. You do learn new skills that way, but it’d just be too much work for now.
Maybe someday. 😃

Hi everyone, can I suggest a feature for a future forum update? Would it be possible to include a translation function? Since I don't speak English, I always have to take screenshots of posts and have them translated by GPT so I can understand what's being

Hakan Önder — Wed, 15 Oct 2025 11:57:19 +0000