Modelling an Internet Market Ecosystem (System Dynamics)

Ross Eyre et al.
Stellenbosch, South Africa
May 2020 
Medium / Processes
Systems Dynamics Modelling
Project Type
Research exercise
Following the System Dynamics methodology, we developed a model of an Internet market ecosystem designed to simulate market dynamics between 'hosts' and 'clients' over a period of two years. Once a base scenario had been simulated, initial conditions were altered in order to generate insights about the behavioural dynamics over time, and to identify key variables & possible leverage points for shifting the system in desired directions.

Model & Base Simulation

Base Scenario
Key model parameters (base scenario):

- Neighbourhood size: 2’000 persons
- Initial hosts: 10
- Initial clients: 0
- Legacy ISP price: R2.50/mbit/Day
- AVG initial host price: R2.00/mbit/Day
Scenario 1 - Initial Host Price
Double initial AVG price hosts charge for services.

Observed Result

The variable is not a high leverage point. The market finds price equilibrium quickly. Given enough competition among hosts, a high initial price (possibly set by first movers) only marginally influences the dynamics over the long term.
Scenario 2 - Host line rental
AVG Line Rental (Host's operating cost)

Blue line (baseline): R0.05/mbit/Day
Red line: R0.1/mbit/Day
Green line: R0.025/mbit/Day

Observed Result

As expected, lowering the cost to hosts associated with line rentals from ISPs reduces the AVG price on the network, and results in a more rapid onboarding of both new hosts and new clients. However, the dynamics of the overall system remain relatively unchanged overall, making this only a minor leverage point.
Scenario 3 - Host initial bandwidth capacity

Double initial bandwidth capacity for hosts.

Blue line (baseline): 100 megabit/host
Red line: 200 megabit/host

Observed Result

Interesting results. The internal price attains a lower value particularly over the first year before it stabilises at day ~500

As expected, fewer hosts can support the network demand due to their larger bandwidth capabilities. This creates larger profits for hosts, as fewer profits are distributed.

The most volatile result of this simulation is the 'Attractiveness to clients' graph, which exhibits a significantly higher attractiveness in the first year when hosts have larger bandwidth capacities.

Initial Analysis

Two initial insights:

1) Competition among hosts drives the AVG network price down over time, reaching parity with host ~operating costs in the base scenario (blue line) at approx. T=450 days
2) From scenario 2: Larger initial bandwidth capacity for hosts leads to significantly higher ‘client attractiveness’ conditions in the first year.