SPEEDEX Transaction Accessibility

[John Wooten]

What: decrease upcoming SPEEDEX transaction fees (perhaps integrate into CAP45).

Why: SPEEDEX offers let validators with reasonably high core counts drastically increase trading throughput, which we think will be the dominant long-term ledger use case.

Inspiration: By decreasing trade execution costs, we can:

• Make markets more accessible to impoverished investors,
• Lower overall spreads via investors making markets without factoring costs to update prices into their profitability outlook (e.x. let them make markets in less liquid assets where they might need 150 updates per 1 trade every 2 hours), and
• Promote smaller lot sizes (including splitting up larger orders across different ledgers) which decrease overall market volatility. [1, 2, 3]

Reasoning: SPEEDEX batch processing happens independent of normal transaction processing in a separate validator SPEEDEX computation each consensus round. Since the validator can use parallelism to drastically increase compute speed for SPEEDEX transactions, they reach much higher throughput than normal offer processing. [4]

In brief summary, SPEEDEX offers:

• Get processed independently of ledger transactions,
• Leverage extremely efficient bulk parallel processing,
• Require less overall validator resource commitments.

As such, we think the fee mechanisms for SPEEDEX offers should be independent of normal transactions and start bidding at 1 stroop.

This lets validators with different core counts process all the transactions they can manage. Consider for instance 70k incoming SPEEDEX offers. An 8-core validator that can only handle 40k offers/round time could take the highest fee ones, whereas another with 24 cores could accept them all at min fee, increasing liquidity and price discovery.

Safety: transaction fees in Stellar are mostly to prevent spam and allow for priority when needed in a congested ledger. Though the exact implementation details are yet to be ironed out, SPEEDEX doesn't require the same spam or priority protections because:

• Trade offers are basically never spam,
• Different validators can focus on order books for different assets, and
• Validators with more cores can handle, at least with current volume, effectively uncapped auction matching for their specific assets each ledger round.
  

As such, it doesn't make sense to apply the current fee structure to SPEEDEX offers.