Blockchain Beyond Transactions

Decentralized apps need modern, data-driven algorithmic decision making

Edward Faulkner
Feb 9, 2018 · 7 min read

Parallelism

Parallelism is a big deal in all of computing. Even the “single-threaded” performance of today’s processors is heavily-dependent on discovering parallel bits of work that can keep your billion transistors busy.² Today’s popular centralized applications take full advantage of parallelism at multiple levels: within a core, across cores, across servers, across data centers.

Node Utilization

Node utilization is the fraction of time that nodes are doing useful work, as opposed to dealing with the overhead of coordinating the network or waiting for input. Current proof-of-work blockchains have approximately zero node utilization, because time spent solving proof-of-work puzzles dominates all other computation, and those puzzles are pure coordination overhead.

Inter-node Redundancy

In the design space of possible distributed systems, we can identify two extreme points. At one extreme — let’s call it max_paranoia— every operation executes on every node. Today's Ethereum operates at max_paranoia.

The Tally Protocol

The Tally Protocol, under active development by Cardstack, is tackling both of the architectural challenges identified above. To give a very brief description: Tally is a Layer 2 consensus algorithm that lets decentralized apps orchestrate staked miners to perform distributed parallel computing on GPU hardware.



Read More

Join our community

To learn more about Cardstack, visit https://cardstack.com.



Cardstack

The experience layer of the decentralized Internet. https://cardstack.com

Edward Faulkner

Written by

Cardstack

Cardstack

The experience layer of the decentralized Internet. https://cardstack.com