Closing Issue 02.
The week opened with the METR teardown. It closes with the two threads still open: chronic sybil contamination in preference data, and the agent decision evaluation vacuum nobody has named yet.
Both solved by the same primitive.
🧵
Issue 02 closes. Five threads, one primitive: human judgement with verifiable uniqueness.
Next week: reward-model QA, benchmark gaming, oversight that actually scales.
If your team is doing the retrofit work, my DMs are open.
The labelling vendor changes. The evaluator does not. The cohort statistic stays meaningful through the switch.
ONTO Wallet is where the durable evaluator anchor actually lives.
Continual training without longitudinal eval is a calibration experiment you cannot read.
The cost surfaces months later as benchmarks that no longer agree.
If your team is building the human-side infrastructure to match, my DMs are open.
Last week's Prism paper treats multimodal continual instruction tuning as the deployed reality.
It also flags that the field is hindered by severe engineering bottlenecks.
The bottlenecks the authors describe are on the model side. The ones on the eval side are larger and quieter.
🧵
Day 2. My AI avatar on the variable every distillation ROI calculation quietly omits, and what preference data integrity actually has to look like.
🎥 ↓
https://t.co/JpIB28K76W
Last week's RTDMD paper proposes reward-guided RL for few-step diffusion alignment.
It also explicitly acknowledges, in its own framing, that aligning distilled models with human preferences remains challenging.
The framework solves a downstream problem. The upstream is still doing what it always did.
🧵
💳 ONTO x @SPACEID is going live.
100 gift cards at $5 each. $500 in gift cards total. No quest, no mint.
For real cross-chain users:
✅ Wallet over 1 year old
✅ Assets across 3+ chains
Connect, authorize, qualify. 7 days. → https://t.co/2PjQnjBanA
Great point. The article touches on this by noting that evaluator credentials can include attestations such as specialist certifications, calibration scores, and inter-rater agreement history. These provide additional context about evaluator qualifications and help auditors assess the reliability of the evaluation process.
The goal of evaluator provenance is not only to make judgments traceable, but also to make the methodology and supporting attestations independently verifiable.
We'd love to hear more of your perspective on this topic. If you're interested, feel free to join the Ontology community, and we'd be happy to welcome you to a future X Space to discuss decentralized identity, evaluator provenance, and trust in AI benchmarks.
The recent debate around the METR benchmark highlights why evaluator provenance matters.
A verifiable chain of evaluators, methodologies, and credentials can help improve transparency, accountability, and trust in benchmark results.
🔹 Decentralized Identifiers (DIDs)
🔹 Verifiable Credentials (VCs)
🔹 Traceable evaluation processes
Read more: https://t.co/PGMgMFs2rE
#Ontology #ONTID #AI #DigitalIdentity #Web3
Vesak lanterns in Sri Lanka for the Vesak Festival.
They are the most iconic visual symbol of the Vesak Festival in Sri Lanka, the holiest Buddhist holiday commemorating the birth, enlightenment, and passing of Lord Buddha.
The publisher does not own the evaluator's record. The evaluator does. The auditor verifies the chain end to end.
ONTO Wallet ships the holder side of that architecture today.
The METR time-horizons graph, cited everywhere from policy briefings to capability roundups, is publicly contested.
A detailed teardown documents "numerous severe errors."
Every lab that ever cited the graph now has a credibility problem they did not have last week.
🧵
🌐 ONTO Wallet now supports more @SPACEID names.
Already in ONTO: .bnb .arb .sol .lens
Newly added: .eth .zkf .manta .gno .cake .burger .wod .floki
One wallet, twelve TLDs, across the chains you actually use.
More tomorrow.
🎮 Ontology x @PalzGame Community Quiz is here!
📅 Friday, May 29
🕘 9AM UTC
📍 Ontology Discord
Special PALZ round this week, so jump into the game first and you'll be ready for the curveballs.
🏆 $ONG prizes up for grabs.
👉 https://t.co/3aa6WNHXgi
🎉 Winners! If we liked or replied to your comment on the competition post, you're in.
DM us your Ontology address before June 4th to claim your prize.
Miss the deadline, miss the reward. ⏰
What if your personal data could finally pay you?
AI learns from the data we create every day - but most users never benefit from it.
With @ONTOWallet + ONT ID, decentralized identity and zkTLS technology are building a future where you stay in control of your digital identity, privacy, and value.
The AI economy is here. Your data matters.
#Ontology #ONTOWallet #ONTID #Web3 #AI #zkTLS #DigitalIdentity
The published research is in.
AI-mediated communication systems measurably shift the opinions of the groups they serve.
Polish, suggest, summarise, rewrite. Each tap nudges. The aggregate shifts.
"Did this person say this thing" is becoming a real question.
🧵 on the architecture that answers it.