So we should put what spare personal bandwidth we have into supporting DIY/community mesh networking initiatives - for all sorts of reasons, this is just one of them, disaster/breakdown resilience is another.
4 Likes
A few positive things with pro farmers is that their uptime will probably be around 100%. Regular people might turn off/sleep their computers several times a day, might be on for a few days and then off for weeks and so on.
5 Likes
Their computers are likely to be on when their “customers” are looking to retrieve data. I know XORspace is very different from physical space but given that chunks will be distributed (in theory) throughout the entire space, I think it is likely that several copies of any chunk will ALWAYS be available even if most of Europe is asleep for example. Or am I wrong?
1 Like
It just seems logical to me that if a high % of farmers have near 100 uptime, that might be something that could possibly benefit the health of the SAFE-network.
I don’t have the tech knowlege to expand further than that. 
2 Likes
Yes this is undeniably true. OTOH if say 75-85% of users have 75-85% uptime and there are 100x as many “home” users as pro farmers then overall the network should be in a pretty healthy state and have fairly short mean lag times for each chunk within a reasonably short period.
Like yourself, this is just a gut feeling and I am not like @neo or @mav and done the sums in any detail.
1 Like
In my thoughts I can see a majority of farmers being pro farmers or enthusiasts running some form dedicated hardware, trying to keep near 100% uptime.
who knows how it will be, I have no clue about what the real distribution might be.
When will sections consider that a farmer is gone and what are the response if certain amount of chunk/copies no longer are online. I was glad to read in a previous update that the team was thinking about how to respond to farmers, downtime and such.
2 Likes
Initially yes, but after some time (bit vague here - 3-6 mnths?) as word of the project gets around, I can see many folk doing a “suck and see” and running a vault themselves having been attracted by the “Dropbox” aspect initially. An awful lot will depend on
-
How we do our marketing
-
The geek reaction to the launch
-
The actual initial rules and rates (T&Cs?) that we define for farming at launch time
-
The economic conditions prevailing at launch time.
I am sure many other factors will play a part as well.
1 Like
For me it all depends on what are the main priorities.
For me those would look something like this:
-
Network health and performane, this priority goes above all other priorities, at all times.
-
Profitable for farmers, this will make sure that farmers try to bring as good bandwidth, compute and storage power as possible, it will also make the network evolve with technology advancements.
-
As many people as possible being able to use the network for free by simply being farmers
My main concern about distribution between farmers is to complete above priorities, after that then other wanted goals could be considered to become priorities, in my opinion.
3 Likes
One more thought from the money point of view. Consider this two farming setups:
- Few years old home or office PC, with 500GB spare space. …something like 75W
- Raspberry-like miniPC with 500GB SSD …less than 5W
With the average price per kW here in Czech Republic running the first one 24/7 just for 1 year will cost me rougly the same as buying and running the second one. Giving new purpose to the old hw sounds nice, and it is a good way to try farming, but I think majority of farmers who willl be in long term will use dedicated hw running 24/7.
4 Likes
One major problem with Bitcoin in my opinion is that the “barrier to entry” have become very high, you need army of ASIC miners and to live in country with extreme cheap electric.
I think and hope that “the barrier to entry” will be much lower on SAFE, some “barrier to entry” might be good. If almost all farmers would need server xeon hardware, running storage from terabytes of ram or similar, that would be something I would not like to see.
2 Likes
This is why I find it interesting to think about and pursue a reward distribution that is theoretically perfectly fair, ie:
For a given period:
my_reward = total_network_reward * my_contribution / total_contribution.
3 Likes
But are there ways to know “total network reward” for the whole SAFE-network. If so, would it be possible to know Safecoins_created or Safecoins_current_in_circulation?
Or would the reward be something similar to:
my_reward = total_section_reward * (my_contribution / total_section_contribution)
total_section_reward = total payed puts to that section for storing
1 Like
Seems there are a few key differences that should make SAFE economy much improved over blockchains. Thanks to everyone for helping distil this out of the fairly complex discussion so far.
Cost profile
In blockchain there is no such thing as ‘spare energy’ that can be put toward mining. Consumers only pay for what they use so that energy cost is critical in determining the amount of benefit from mining.
In SAFE many people pay for internet bandwidth that ends up being unused. This means it makes sense to put that unused resource to work by running a node. Any reward can be used to offset a cost that would have to be paid even in the absence of running a node.
This means it’s more likely that professional operators can function alongside hobby operations.
Granularity
It makes no sense to mine tiny amounts of bitcoin since they can’t be transferred due to dust limit and high fee restrictions.
However SAFE is expected to have much lower fees and finer degree of granularity so small amounts of work are still useful and able to be rewarded. This should encourage small players to feel it’s worth taking part.
Suppliers
The supply of energy that fuels blockchain projects is very different to the supply of bandwidth which fuels SAFE economy.
- Energy generation can be owned by the miner, whereas bandwidth is generally purchased from a supplier and cannot be privately owned or generated.
- Energy is pay for what you use, whereas bandwidth is usually paid for monthly with any unused amount forfeited.
- Energy can operate in isolation, but bandwidth is a network of related parts and is difficult to operate without forming business relationships and integration. Energy is operational, bandwidth is cooperational.
- Most energy has significant ongoing costs (ie the fuel to generate electricity, renewables are the exception) but bandwidth is almost entirely upfront capital with relatively small ongoing costs.
It’s not clear to me yet the particular impact of these differences but it’s certain to show itself in the economics.
Custom Hardware
Bitcoin saw many orders of magnitude improvement in energy efficiency with the introduction of ASIC which made any miner without that hardware essentially uncompetitive and stop mining. This put the decentralization of bitcoin at risk.
That degree of improvement is not possible for bandwidth efficiency. This means that new operations should not drive out existing operators to such a significant degree, improving the rates of participation and degree of decentralization.
Does this sound right? Did I miss anything?
@VaCrunch “factor of randomness” I think the randomness idea is interesting. It really ties in with human psychology, gambling, rewards, anticipation, risk etc. I don’t think it helps from a mathematical or idealistic perspective, but the human aspect of the economy is very interesting to look at. There’s no denying that (at least initially) the lottery aspect of bitcoin rewards was part of what made mining fun and interesting. But then the lottery aspect added too much risk and pools became essential.
@Southside “supporting DIY/community mesh networking initiatives” The impact of let’s call them public networks (as opposed to corporate ISPs) in the future will be really interesting. Mesh networks etc will hopefully make SAFE even more resilient. I hear stories about how much of a problem comcast is in USA, it makes me wonder about viability in general of a bandwidth-based economy in the current world we live in and how things might need to change in the future for a bandwidth based economy to be sustainable.
10 Likes
One thing I have been thinking about lately that came to mind when reading your comment, is the difference between SAFE and Bitcoin reward.
A Bitcoin miner can’t live from transaction fees, a small $1 transaction is the same cost as $100 000 transaction (with equal size in kilobyte, everything else equal).
That makes it so Bitcoin miners need to earn profit from the new coins that are created when a new block has been mined.
SAFE farmers can make profit from “cost of storage” alone and don’t need new coins to be created to earn profit. Paying for 1mb or 1000mb scales linear, in difference to “Bitcoin transactions cost” which is flat at any given rate.
Do the SAFE-network need to create new coins? With a supply/demand Safecoin market and a supply/demand farmer/client market, the whole system would find equlibrium.
If the amount of coins always is known then that would create a well functioning market for Safecoins.
A fear is having market where current/total supply would say:
?/4 500 000 000
that would create some bad effects.
5 Likes
I am thinking that there will be an economy of people building vaults into SBCs and selling them for a reasonable price. Easier to setup than any router for the home user and can be farming within minutes as long as the person already has an account, and take 10 minutes if they need to setup an account.
This solution minimises electricity to like less than 10 watts, more like 5 watts. 2 models - Mag drive or SSD with the mag drive the cheaper option and suffers 5 to 10% slower overall in the round trip.
This will likely be cheaper for the home user who only has their PC on for limited time (say a few hours) each day. Some have their PCs on for 16 to 24 hours per day and the extra time to make up 24 hours is possibly cheaper than the couple hundred for a fast SBC unit.
I still believe we need to be rewarding all who have successfully held the requested chunk and supplies the signature in reasonable time, because they all have in good faith done what is needed and are benefiting the network by doing that
This also aids in reducing the benefits of expensive hardware/datacentre just to gain rewards using that slight advantage. And in the long run that would drive up the actual costs to store data due to “arms race” if more powerful equipment has significant benefit
11 Likes
Just popping in … may not read replies, so take this for what it’s worth.
Aside from crypto-currencies which are self-regulating through code that everyone can see, the general marketplace has many hidden politicized factors. E.g. Jeff Bezos goes to washington to complain that he’s not being given a fair shake from mom and pop on main street, so write a law to make it fair for me and by the way here’s some money for your re-election campaign.
We all sense that existing politicized markets are not ‘fair’, because we know that people use power and influence to bias things in their favor. But with crypto and also with the Safe Network, we have proof through code. As such I think we should not worry about centralization. As we know, centralization can bring efficiencies of scale - which is good for the network. So long as the network can respond to bad actors, then that should be fine as even the little guy is going to benefit from it. Too often the word ‘fair’ is abused to represent one’s own subjective viewpoint. The network needs to be neutral to be successful and incorporating subjective viewpoints on fairness is troubling.
On another point, in dealing with storage of low-popularity data, I wonder if the network could ‘bill’ a farmer for not having what they should have. e.g. a request for a chunk is rejected by a farmer who should have it, so instead of paying a reward, a bill is issued instead. So carrot and stick, not merely carrot. The bill would likely come off future rewards since ripping it out of their account should be impossible. Taken a step further if there is some metric of popularity (I assume there is as the network increases copies for popularity?) then adjust the reward/bill appropriately. Or just use reward mechanism, but adjust reward for popularity - more popular less reward.
Thanks for reading and thanks for the discussion. Most interesting.
2 Likes
May not be particularly useful or insightful but this reminded me of (idealised) competitive ideal free foraging that I read about eons ago in pigeons and cooperative foraging more generally. In one study on Starlings dominant individuals (In our case well bandwidth resourced farmers for lack of a better name) were less able to defend their competitive advantage against groups of subordinates over a certain number. So along these lines in our case would perhaps be similar to a bittorrent-like sub-chunk swarm delivery by bandwidth reduced farmers being able to challenge much better bandwidth resourced farmers for a stable competitive system.
4 Likes
Small miners often (usually) mine as speculation because they believe the price will go up 10x, 100x, or even 1000x plus in the future. In my experience, much (if not most) smalltime mining is actually unprofitable or only marginally profitable based on current prices. Many people have mined altcoins from day one, even when there is no existing market, and only a dim hope for one in the future. Some do so ideologically, others purely as speculation, some a mix. And this is a rational thing to do, so long as there is an expectation the coin value will likely increase in the future. A small loss/investment today can lead to relatively huge gains later on. So I think that a simplistic cost/benefit analysis based on current prices misses a big piece of the picture with respect to the long tail of contributors and their motivations/behavior.
8 Likes
I agree with this. Considering the utility of the network it seems like users will farm for protecting network data (continued access to the data is their main reward, farming protects that access and safecoin is a side effect).
For example archive.org had 11.5M USD contributions (see p9) in 2018, at average $45 per contribution that’s about 250K donors. I know giving cash is not the same as starting a vault, but this gives me reason to think there’s a lot of people who will care about the non-farming aspects of running a vault.
10 Likes
How about a new request type that only vaults can make (ie not clients): starting at <xorname> return the first N chunk names you have stored
This would allow vaults to prefill their cache (by requesting chunks by name like a normal client would) and also allow some degree of auditing since multiple vaults would overlap the name range and should all respond the same way.
It’d be cool to be able to do audits without impacting the workload of elders.
3 Likes