Real World Ethereum 🌎 
building /daimo ✳️ 

Building @farquest, wield.xyz, investing @ miy.com. ex-Meta, ex-Coinbase // cryptojcdenton.twitter // FID #251

prev american energy data and cofounded rocketr.net swish.id 

BDFL of Quilibrium, Eng @ Farcaster, ex-Coinbase, always opinionated, never hydrated

/quilibrium

Working on a ML model which will output an 'address score' similar to how credit scores work (~low is 300-500, mid 500-650, high 650+)

some interesting data:

me: 529.52
@alexpaden: 567.79
@cassie: 685.06
@jc: 698.19
@dwr.eth: 663.29
@dcposch.eth: 752.81

building far.quest & /farmarket. cofounder at miy.com ventures. FID #274. Previously Square/Cash App. hi_im_nico.twitter

I am planning to open-source the model. My training data and scoring system depend on a dozens of factor, notably num_transactions_score, transaction_frequency, total_value

Goal: decentralized consumer reports anyone?

Just if it's not clear: this pre-trained model will be open-sourced.  

I am training with FC addresses to start, and trying to get an idea on the scoring scale - are these too low? too high?

I actively wanted to avoid this hence trying to gauge feedback before the training begins

Are the scores in general too 'low'?

tfw my social credit score is lower than my credit score https://i.imgur.com/d2T303O.jpg

We make crypto exploration fun through far.quest 👁️

See wield.co for more!

As promised, heres a link to our open-source tensorflow scoring service on a @beb dimension, hardcoded to an address for now.

https://github.com/bebverse/universe/blob/main/services/ScoreService.js

Try the endpoint at https://universe.b5b.xyz/score/0x0 to see tensorflow on node.js in production!

Stay tuned!

🎩🍌Ethereum builder & people connector. @juicebox revnet.eth.limo defifa.net d33m:liv

if i lead the top score on castrater then it's just cosmic irony

Can I post the page and the open-sourced model once it's done?

👏 I have castrater.xyz for promising FOSS projects. This sounds promising! All said assuming my credit score is f'n superior.

The only problem is externality pricing.

Building https://negationgame.com

seemed like an widely accepted but imperfect system. The training data and weights are based off the scoring criteria

Building physical commerce to give artists and writers other ways to monetize their work /clickcast /cpg

@pushix mentioned https://roci.fi/ for credit score. 

I didn't mean mean this to be a defi lending score - the idea was more a 'social score'

I’ve thought about this, but it’s awkward to pin it to existing priors like FICO. Becomes interesting as more lending data is available on chain I’d think

Founder https://axflow.dev // creator of the https://blockchainsmokers.xyz // nf.td/nicho

🌶️ takes: any computed ML score which depend solely on on-chain data will be flawed. Also: any centralized score which incorporate only a set of external APIs (e.g FC but not Lens) is flawed

We need a decentralized 'karma' system where each company can report their own

product @ grayscale.com • all povs = my own

I hear you, I have worked in fintech most of my life and agree FICO score can be very biased

The score depends on the sum of weights and normalization of dozens of on-chain criteria.

My thoughts on eliminating biases: we need a decentralized scoring system where different identities can report their own member score.

Would it also have the same bias that systematically under-represents minority groups like FICO?

Building /bright-moments - minting onchain art IRL. https://brightmoments.io | /purple #15 | FID 129 #magikarp

732.514921556241 for 0x1040259f6Fe4f1C6ed1044b9Df6a8A4eBD2d3511.

One of the highest!

VC & Engineer | FOGC | Researching, Investing & Building — robinnordnes.com & elixircapital.xyz

Yes. The pre-trained model will be open-sourced. Anyone can host it. 

However: this should not be the final score. To eliminate biases, we need different reporting entities - e.g. Farcaster give you a +20 if you are an active caster. 

The final score can be computed by any aggregators when everything is open-sourced

Would be really cool to see how you calculate the scores! Do you plan to open source this?

Principal at Cobalt Ventures. Write @ TheItinerant.io ; BenErsing.com           Host: /venturecapital

The scores depend on a dozens of factors like transaction frequency, age of address, total value. It is more like a 'reputation' score than a credit score!

Interesting concept. What would be my score? Actual credit score is 780+

Llanfairpwllgwyngyllgogerychwyrndrobwllllantysiliogogogoch, Llanfairpwllgwyngyll, UK

592.06 for 0x466928b1e8c757fA1e94Db11209DdCf28F86E37a

I would like to know mine please. praven.ens

697.52 for 0xE7657Ab31a15A0A2E024bA12ab85967FEC4253bc

Web3 email and messaging for everyone. E2EE.
Builders, get the SDK: https://docs.mailchain.com/developer

https://mailchain.com

I heard good things about mailchain from @stoic, this could really be useful in cold mailing someone (vet addresses, send tokens to message higher signal addresses, etc)

Fellow startup founder ◡ film producer.

High on bleeding-edge ideas & tech.

nf.td/jithinraj

0x09dE8dE37fD4BEfE77a45EDa7E794A2bEbBDC1B7 has 563.97

Hey @n  Would like to know mine. jithinraj.eth

Thank you

working on /receipts (receipts.xyz) // nf.td/victor

Thanks Victor! I think there's a good opportunity in eliminate traditional centralized score biases (working in fintech the assessment can be pretty inaccurate)

e.g. if any entities (like NFTD) can have a customized scoring algorithm the final score will be balanced assessment of an individual's reputation.

love that you’re tackling this nico, and I think framing it on a scale that’s already familiar is smart

The model is trained a dozens of on-chain data, like unique counterparties, total tx value, tx amount

Reseach & Investments at L2IV // arhat.mirror.xyz // 0xArhat.twitter

it seems like the address is a brand new without any history on ETH? the model isn't trained on Polygon/Matic where I see you have the majority of your activities

I mean, it will technically work on any EVM chains. I had more thoughts on Polygon/L2s and think there should be different scoring system due to gas differences

Fwiw the original model has some weight for total gas spent (ETH)

what's mine? 0x34ddd9223D9DDb6B56F640824Af6FCC31e1deBF4