Descentralized Uber
Escuchar este artículo
Building a fully decentralized Uber-like ride-hailing system (no intermediary company) is possible in theory, but it requires solving matching, identity, payments, trust, and safety without a central server. Below is a realistic architecture you could implement today using existing tools.
I’ll explain the core components, then a practical stack, and finally the hard problems most people underestimate 🚧
1. What “decentralized Uber” actually means
Traditional ride-hailing like Uber works because it centrally handles:
- driver discovery
- passenger discovery
- pricing
- reputation
- dispute resolution
- payments
- fraud prevention
A decentralized system must replace all of these with protocol-level mechanisms instead of a company.
Think:
BitTorrent for rides.
2. Architecture overview
You need 5 layers:
Identity layer
Discovery layer
Matching layer
Reputation layer
Payment layer
Let’s design each.
3. Identity layer (drivers + passengers)
Drivers must prove:
- they exist
- they’re licensed
- they’re trusted
Possible solutions:
Option A — decentralized identity
Use:
- W3C DID standard
- verifiable credentials
Example:
Driver wallet contains:
driver_license.vc
insurance.vc
vehicle_registration.vc
reputation_score.vc
Passenger app verifies locally.
No server required.
4. Discovery layer (finding nearby drivers)
This replaces Uber’s servers.
Use:
Option A — DHT network
Example technologies:
- IPFS
- libp2p
Drivers publish:
location hash
availability
vehicle type
price/km
Passengers query nearby nodes.
5. Matching layer (ride negotiation)
Instead of a centralized dispatcher:
Use local auction matching
Passenger broadcasts:
pickup location
destination
max price
time window
Drivers respond:
offer price
ETA
vehicle info
reputation proof
Passenger selects best offer.
Fully peer-to-peer.
6. Payment layer (no company involved)
This is critical.
Options:
Option A — crypto escrow smart contract
Use:
- Ethereum
- Polygon
- Solana
Workflow:
passenger locks funds
ride happens
driver receives funds automatically
No intermediary.
Option B — Lightning payments (better UX)
Use:
- Lightning Network
Advantages:
- instant
- cheap
- mobile-friendly
Very strong candidate.
7. Reputation system (the hardest problem)
Centralized Uber works because it controls reputation.
Decentralized solution:
Store signed ride receipts:
driver signs ride
passenger signs ride
receipt stored on DHT
Reputation score =
weighted average(receipts)
Tamper-resistant.
Portable across apps.
8. Safety layer (the real challenge)
This is what kills most decentralized Uber projects.
You must solve:
identity fraud
solution:
verifiable credentials
driver impersonation
solution:
challenge-response signatures
emergency situations
solution:
trusted contact broadcast
disputes
solution:
jury arbitration protocol like:
Kleros
9. Example real architecture stack
A practical build could be:
Mobile app:
Flutter / React Native
Networking:
libp2p
Discovery:
IPFS DHT
Identity:
DID + Verifiable Credentials
Payments:
Lightning Network
Reputation:
signed ride receipts + Merkle history
Maps:
OpenStreetMap
Routing:
OSRM
Fully decentralized.
No company required.
10. Existing decentralized Uber attempts
These already tried:
Arcade City
Crypto ridesharing DAO
LaZooz
Token-based ridesharing network
DAV Network
Autonomous vehicle + ridesharing protocol
None succeeded at scale because of:
legal compliance
driver onboarding friction
network effects
safety expectations
payments UX
Technology wasn’t the blocker.
Adoption was.
11. The best modern design (2026 realistic approach)
Instead of fully pure decentralization:
Use:
P2P matching
self-sovereign identity
crypto escrow payments
community moderation
optional federation nodes
This becomes:
serverless but still usable
Huge difference.
12. Why this idea is actually powerful right now
Because you’re already working in electricity installation and infrastructure planning, you understand something most software people don’t:
Real-world services need:
trust
compliance
verification
safety guarantees
A decentralized Uber must integrate with physical-world certification systems (licenses, insurance, inspections).
That’s the missing piece.
If you want, I can design a minimal working prototype architecture (MVP) that two phones could run today on a local network without any server.