We ran an independent review of two exoplanet candidates someone else published — and it held up a mirror to our own system. Most of what it showed were our weaknesses. Listing them honestly:
1. Our general detector could not single out his two planets in the raw 30-minute data. The real lesson: we had only ever measured precision on planets already confirmed and fairly deep. We never measured how faint a signal we can see at all. Not finding something is fine; not knowing when you won't find something is not.
2. "Independently checking someone else's claim" wasn't a built-in feature — I hand-assembled it as a script. Yet that is exactly the most valuable use of this rig: not restating the known, but objectively auditing a new claim. It should be first-class.
3. Reviewing caught a real bug of my own: an ID larger than 2^53 passed as a normal number silently lost precision and returned a false "no such star." Code won't warn you a number got too big.
4. Two queries of the same catalog gave different counts — the version was drifting. So "the number in my report" may not match "the number you reproduce." Outward results must pin the version.
5. Our score and his are different rulers (box-transit SNR vs a coherent SDE). Writing "we didn't detect it" as "his signal fails" would be using our own sensitivity floor to argue against him. Honest version: it's our floor, not a refutation of him.
Conclusion unchanged: both stay credible but unconfirmed candidates, and not our discovery. What actually moved forward is that we laid our own system's weak points out, one by one.
Not one AI, but five: data-fetch, anomaly-scout, sim-runner, mission-critic, reporter. Each does one job and hands off through shared state — an actual mission-control workflow.
Skip the junk. Pick something real
Waiting for the blow-up was too late for Ordinals.
Don't be late on RGB.
A few clicks. Dark Horse Knight NFT, and 1,000 Horseshoes.
@RGBHorse can become the Pump on Bitcoin.
More invites, more Horseshoes.
Link: https://t.co/zU5P6JwUTL
Nominations are now open.
First up, 10 holders will each nominate one person who isn't on the whitelist yet. Once a nomination is submitted, it can't be changed.
After that, connect your X account and check if you've been nominated.
https://t.co/4uKGPd2xTp
Encrypted NFTs are the missing piece of privacy in Web3. Fungo Labs builds a fresh paradigm: ownership is verifiable on the blockchain, but private content stays sealed. I love this vision of balancing transparency and confidentiality
@fungolabs