Keeping Safecoin Decentralised

Vigilant readers of this blog will have noticed that MaidSafe is very focussed on decentralisation :). So much so in fact that we don’t use a blockchain for consensus and we are setting up remote developer pods to compete with us to maintain the underlying network code. It would therefore be unfortunate if safecoin farming became centralised, putting a large proportion of safecoin in the hands of just a few large groups. While other projects have attempted to tackle this issue technically, MaidSafe has taken a different approach and focussed on economics.

The following high level overview is taken from a section of the safecoin paper that we published a few weeks back. The reason for regurgitating it here is because not everyone likes to read papers and this is an area that many within the community have shown a particular interest in. While this information does not get us much closer to predicting how many safecoin you will earn for providing X resource, it does convey the over arching intention of the design.

As figure 1 demonstrates, the safecoin earning algorithm is based on a Sigmoid curve, in that all vaults earn, slowly at first and the rate increases as the farmer stores up to the network average. The earning rate also takes into account the rank of the vault, a process whereby the network scores the usefulness of each node from 0 (being the worst) to 1 (the best). The safecoin farming rate is ultimately the result of the network rate, a balance of the demand and supply on the network, multiplied by the vault rank.

safecoin farming speed

The network rate will start to level at 20% above average, thus discouraging massive vaults which would bring centralisation to the network’s farming process. Safecoin is allocated to them by the network and is paid to the successful node as data is retrieved from it (GETS), as opposed to when it is stored (PUTS).

The network automatically increases farming rewards as space is required and reduces them as space becomes abundant. Data is evenly distributed on the network and therefore farmers looking to maximise their earnings may do so by running several average performance nodes rather than one high specification node.

One comment

  1. The article describes the key to keeping Safecoin farming and distribution decentralised – the economic model. I didn’t really get the significance of this until I learned how the network works, and began to understand how it favours small scale farmers, so I want to explain how I understand this in case it helps others.

    Firstly, the economic formula alone doesn’t make it uneconomic to farm at scale, at least I don’t see it that easy. I think it does make it much harder to corner all the rewards, and to prevent other farmers being able to farm by making small scale farming uneconomic. But on its own, it would still mean that putting more resources into farming increases rewards roughly in proportion to infrastructure investment, with economies of scale still favoring larger operations. However, SAFE Network as a system does disincentivise this, and puts a fairly low ceiling on large scale farms (many vaults operated by one entity) or farming pools.

    This is a consequence of two factors:
    1) Anyone can farm effectively. Being big doesn’t give you the ability to hoover up all the stored data, because of the points made in the article. This is different from bitcoin, where the key to farming, or rather mining, is computational superiority, and the more money thrown at this, the more you can dominate mining. With Safecoin farming, the sigmoid distribution caps this at 20% above what average person with a quality PC and internet connection (network average) can achieve, which is what ensures they can’t be shut out. But the second factor is also very important IMO…

    2) Small farmers have spare resources available at zero cost, while a dedicated farming operation has real costs at any scale. If hardware is dedicated to farming, all those costs must be met out of farming rewards. Whereas anyone who is using an existing PC, tablet etc. for other things, has already discounted these costs, and they have unused resources available to share with the network at little or no additional cost. They can farm at any level and be profitable. The effect of this is to ensure the network will always have access to very low cost resources. Providing the are enough such resources, they create a downward pressure in the amount of Safecoin rewards it needs to offer to farmers, which tends to negate the relatively marginal improvements in profitability from economies of scale.

    The second factor doesn’t eliminate economies of scale, but it shifts the point at which large scale farming hits a profitability ceiling, and the more zero cost farmers participation in the network, the less profitable large scale farming becomes.

    So, will there be enough small farmers for this to be effective? I don’t know, but it seems likely.

    I think we can expect many, many, SAFE Network users who will want Safecoin to pay for services such as storage, and will find it easier and more attractive to farm what they need than to purchase it with money. The number of small farmers with negligible costs will then both be large, and scale with the network.

    So SAFE Network discourages centralisation of farming by limiting the profitability of scale, and avoids the centralisation risks of a computational “proof of work” approach, that we’ve seen is vulnerable to investment in technology such as ASIC mining processors.

Comments are closed.