Potential misconceptions with the safecoin economy

Autonomi Network Token (incl (e)MAID)
22

For me, this is the nub of the question. I am almost always in favour of repurposing old kit, using what we already have in new and more efficient ways rather than buying new kit.

Analogy, I’d rather run my old diesel car on filtered chip-fat - cos it’s cheap, I dispose of “stuff” that would otherwise go to waste and I already have a car. I dont have to buy an expensive new car and the pollution “burden” is waaaay offset against the pollution burden of building a new car, electric or otherwise.

So I am very much in favour of using what we already have - This of course can only go so far. As above if everyone copied my example, the price of used chip fat would soar, chippies would make more money giving away fish suppers and we be sick of the smell of chips.

So we need a balance and we need some mechanism to “pull the levers”

3 Likes
23

I do believe that one way of rewarding vaults will allow those who seek better lag times (hard on a global basis since light only travels so fast) and those who wish to minimise cost and use their home setup (or a SBC + Disk)

That is to reward all the vaults that successfully prove they had the chunk. But the vault that supplies the chunk the fastest (here global lag times rule) gets a greater proportion (maybe half) of the total farming reward.

This is a major difference between miners and farmers. The miners are CPU/GPU/ASIC (computing) intensive and farming is disk access to network. For the miners they spend a lot of time computing a block (real long time), but the farmers are accessing disk (sub 100mS) and then sending it out on the network where global lag times rule.

Compare the average block compute time to the average disk access to Internet time and therein lies a major part of the equation.

Our thinking in terms of mining will always lead us astray when considering what is best for SAFE. While considering mining can help in some ways it will also hinder us in others

8 Likes
24

There are 3 major areas where improvements can be made for a vault performance

  1. Disk access speed. (IOPS is the important figure, but this is a rough consideration) Total time to access and read a random 1MB chunk
    Recent (say after 2017 with decent mag drive) Home PC - sub 50mS average
    SBC with SSD say 15mS
    Home/small business NAS direct connect to computer (or SAFE running on NAS) say 30-40mS since most NAS are magnetic media. The speed of NAS comes from multiple accesses to the NAS and handling that faster than a single drive
    Data Centre (iSCSI or similar say) 30 mSec (still looking at single chunk access on mag drives)

  2. Internet Speed (bps and lag time to get to point of presence of ISP/data centre
    Home on the order of 25mSec and 100 Mbps (assuming 2 years at least for significant adoption by public).
    Data Centre 10mS and 1000 Mbps

  3. CPU Speed
    All x86 processors of the last 5 years will provide similar performance for running SAFE
    Phones/ARM processors of last couple years and choosing a good performance one will give similar results to PC
    The reason is that SAFE vault process is not heavy on the processor (when SAFE is optimised and not debuging) At this time though this is not the case and could see differences between PC and ARM. Phones is more unknown

For a vault to successfully deliver a chunk,

  • the section has to request the vaults holding the chunk to retrieve the chunk.
  • The vaults then retrieve the chunk
  • The vaults respond with signature of the chunk to validate they have a valid copy of the chunk. Small packet of Data.
  • I assume the vault that responded the fastest then retrieves the chunk to the network.

Now at each stage where communications to/from any of the vaults is concerned, the time for that communications is dependent on two factors
a) the precise sequence of events leading up to it since that will determine which elder(s) contact which vault
b) The lag time between the elder and the vault. This will vary considerably and on average be measured in terms of 100 to 500 mSecs

Now with computer time being a fairly constant factor no matter what system is used since it is extremely small for a vault access we can effectively discount it in comparisons for home system to high performance vault.

Now since the packets requesting the chunk and the response with signature are small we can discount the bps since it takes 0.04mSec to send at 100Mbps

Thus for 3 example setups we can approximate the time to send the response packet with signature to the point of presence for ISP/data centre as

  • Home 50+25 = 75 mSec
  • SBC (Home) 15+25 = 40 mSec (SSD)
  • Data Centre 30+10 = 40mSec

Now remember that the time from request for chunk to delivery of chunk signature packet is what will determine which vault sends the chunk and is rewarded. This requires a 2 times lag from elder to vault.

Various cases

  • Elder sending/receiving the request/sig is on the same network - only expect this if both in data centre
    • data centre computer (obviously) at 30 mSec
  • Elder within 1000 KM (approx 15mSec lag time)
    • Home (slower drive) 75+30 = 105 mSec
    • SBC+SSD 40+30 = 70 mSec
    • DC 40+30 mSec = 70 mSec
  • Elder Across the USA or Australia 5000 KM (approx 75mSec lag time)
    • Home (slower drive 75+150 = 225 mSec
    • SBC+SSD 40+150 = 190 mSec
    • DC 40+150 mSec = 190 mSec
  • Elder 10000 KM (approx 150 mSec lag)
    • Home 375 mSec
    • SBC 340 mSec
    • DC 340 mSec

Those figures are based on best times through the network routers, and there can be delays and re-transmissions of packets causing some times to increase. Also very approx lag times used for illustration and not meant to be an exact figure. Its the approx differences that is important.

From the figures we can see that improving the drive can help the response be marginally faster. But in the real network we will have vaults around the world so the elder to vault distance will vary depending on the location of the vault for a particular chunk request.

For instance if the home PC with the slower drive is 100 KM from the elder and the DC or SSD computer is 1000 KM away then elder will receive the signature packet at around the same time.

Now for the globe the variation is up to 20,000 KM and this would mean even making the vault have a zero time to get the signature returned would only have on the order of 10% better chance of beating a reasonable home computer on average.

10 Likes
25

What this suggests I think, is that if it were cost effective to locate vaults around the major areas of demand, it would tend to centralise both economically (for those with the resources to do this) and to a degree geographically.

However, this will be balance by the competition between those able to do this, both with each other and with all the vaults which just happen to be close by. I think it is in this scenario, where a few ms shaved of a disk access etc would have most impact.

Such clustering will happen, but will be hard to model because other factors will come into play such as increasing overheads in areas of more demand (due to higher living costs).

7 Likes
26

Where larger groups of people live will be where the clusters are. And they usually have better Internet connections. So there will be a natural occurrence of this.

BUT you have to remember that sections are not based on geo locations, so even if every big town and city had these clusters (they will anyhow since the people are there) the lag times will still be global. The clustering will really have little effect since there will be 10’s or 100’s of thousand such clusters. The elders of a sections will be spread across all these areas and then also in non-clustered areas.

I doubt clustering will see much of a difference in the lag times. Maybe for the 100 big cities around the world you will see a significant amount sections with 3 elders from one section in the same city. Still a small number in absolute and relative terms.

Then you have no choice in which sections that will be for and you have no choice which section your vault will be in.

My take from that back of envelope analysis is that spending 1000’s of dollars to etch out that 10 mSec extra speed by having sub 20 mSec total response time (request to signature reaching point of presence) is not going to gain you much. It might take years to recover those 1000’s if you ever do before needing to replace the expensive H/W

And then a decent SBC costing less than 100 and a SSD costing a couple of hundred will be so close and recover its costs in a fraction of the time.

Whereas mining you spend a few hundred on an ASIC and get 100’s or 1000’s times more than a more expensive graphics card.

The scale of improvement from a vanilla home system or a better a SBC+SSD and a very expensive setup to get lower drive access times, is not there for farming. Different set of parameters

4 Likes
27

I personally desire a fair farming system that allows all people to participate in farming on a relatively level footing. I want to see unused resources put to work. I’m really interested in doing better than blockchain and proof-of-work.

But despite that I’m going to dig the hole a bit deeper here against my own interests. I’m finding it hard to see the ideal scenario playing out in reality, and no matter how we approach the economic algorithm it will probably struggle to ‘enforce fairness’. Fairness is not an emergent property of a competitive system (which the economic algorithm must be, or so it seems to me).

From the sia mining article linked above:

“The ASIC game has become such an advanced game because there is so much money on the table. Even small coins can be worth tens of millions of dollars, which is more than enough to justify a high-risk production run.” (feel free to replace ASIC with purpose-built hardware in the case of SAFE).

MaidSafeCoin is already well across that ten million dollars threshold. Imagine the value on the table when the network is functional and using real safecoin. Will we really be able to keep everyone in the farming game? Even if it’s not ASIC as the target of production it will be whatever the bottleneck is, and a lot of people will be producing to out-compete others.

I hope it isn’t like that, but from the existing evidence I see in decentralized projects this will be very difficult to avoid.

Looking at stats from burstcoin, a proof-of-storage based blockchain, it’s kinda staggering how far beyond consumer-level let’s-just-have-a-go it is. The middle ground for being considered a small miner is 10 TB (see small miner pool stats below). That’s huge!!

Burstcoin is #339 on coinmarketcap with a market cap of $11M.

Total network size is about 300 PB - https://explore.burst.cryptoguru.org/chart/supply/network_size

Using the cryptoguru pool stats below, that pool accounts for about 17 PB or about 6% of the network total.

From this you could extrapolate and say there’s about 6500 miners on the network (284+96+14)/0.06

“Small miner” pool

https://0-100-pool.burst.cryptoguru.org/

Looking at the All Miners tab

284 miners

Median capacity: 9.815 TB

Avg capacity: 24.971

Total capacity: 7066 TB

Medium miner pool

https://50-50-pool.burst.cryptoguru.org/

96 miners

Median capacity: 36 TB

Avg capacity: 55 TB

Total capacity: 5236 TB

Large miner pool

https://100-0-pool.burst.cryptoguru.org/

14 miners

Median capacity: 434 TB

Avg capacity: 419 TB

Total capacity: 5454 TB

To be fair to the SAFE network it’s not quite right to directly compare these storage figures since burstcoin storage is done by generating offline chunks whereas SAFE storage is done by downloading chunks, so will be constrained by bandwidth rather than storage.

Might be worth asking when comparing burstcoin and SAFE, is bandwidth a more openly and fairly distributed resource than disk space? Is bandwidth going to give a more fair mechanic than using storage? Disk storage can be privately purchased and owned and operated, but bandwidth cannot. I would guess that bandwidth being the bottleneck probably makes the network more at the mercy of ISPs and external authorities than if it were purely storage based.

Just stirring the pot here, I want to reiterate my personal goals are to arrive at a more efficient and fair system than the blockchains have done.

@VaCrunch I think you’re referring to the sigmoid concept, I can’t find it now but I do remember this concept myself too.

@happybeing, you have the right idea of my intentions at the personal level. But at the analytical level I can’t yet find a way to believe that industry won’t dominate SAFE farming. As you have identified, rewarding participation rather than performance is probably a pretty good heuristic to start from.

@southside you put it nicely, it would be great to be able to get the most out of what we already have. But I feel there’s a balance between using the current tech and pushing toward better new tech. Hard to say where the balance lies.

@neo really appreciate the analysis. Interesting idea about portioning the reward.

“the vault that supplies the chunk the fastest”.
Would love to hear more about this. Especially some details about what ‘fastest’ means; fastest to the client, or to the elders, or what… It’s not clear to me how to measure fastest, but it could be handy to have it.

“Compare the average block compute time to the average disk access to Internet time and therein lies a major part of the equation.”
This sounds a bit mixed up to me. Disk access time is more comparable to hash time in bitcoin, not block time. Maybe you could explain this comparison a bit more? The way I would differentiate this is one is a ‘unit of work’ (hash / chunk access) and the other is a ‘unit of reward’ (block / reward payment). The unit of work can be optimised by the miner/farmer, the unit of reward is the lever of control for the network.

“There are 3 major areas where improvements can be made for a vault performance”
This is a really nice compilation and analysis. It gives me reasonable hope that whatever competitive edges are discovered in the future won’t push out smaller farmers. I wonder how high frequency traders would compare with these figures. I think the HFT mob are probably the best point of comparison rather than retail or consumer level networking.

13 Likes
28

As things stand, it’s your elders that’d provide the reward, so fastest to elders would be simplest. (That doesn’t mean that can’t be changed, but just FYI…).

I think fastest to client would be impractical in that’d you’d have to trust the client to be honest about what it received/when. Especially when GET access is free of cost or consequence.

10 Likes
29

So this seems to be generally true in the cryptocurrency space today. However, my observations are that:

  1. In the ideal system, each participant would be rewarded in exact proportion to their useful work/resources contributed to the system. If such can be technically achieved, then maximum decentralization is strongly incentivized. ie, imagine if every cell phone could mine bitcoin while charging and receive satoshis equivalent to the hashes performed, no matter how small. They are not competing with the big guys but rather participating or helping. But they can’t because…

  2. The all-or-nothing block reward gets paid to a single “winner”. So an ideal mining/reward system would pay out to all participants according to their share of work. Mining pools approach this ideal (as an ugly, centralized add-on), but there is still the bigger problem that…

  3. The present (not necessarily future) “real world” value of the work performed is too small to even justify network fees of sending the payment, and in bitcoin could be even less than the “dust” threshold (itself an ugly hack). The fees are primarily a limitation imposed by the inefficient global consensus/order requirement of the blockchain model. So this makes it basically impossible for the little guy to even participate, which has knock-on effects that new money is less widely distributed, fewer people interested to participate, greater inequality, etc, etc.

This is actually a huge issue for bitcoin et al, truly an existential challenge over the long term, given that it relies on wide decentralization to enforce the consensus rules. The long tail of tiny participants would make the system so much stronger against consensus attacks.

Now in the AT2 system, the real cost of consensus goes way down. Also SafeCoin has a stated goal of fee-less tx, like cash. If this can be achieved, and there is sufficient divisibility of the currency, then this opens the door for much much broader decentralization, as tiny payments can be made for tiny contributions. Old/weak devices may participate, rather than compete.

or so I’ve thought for a while…

19 Likes
30

Good points @danda, I like!

4 Likes
31

Well said. You articulated the key nuance I’ve been considering on this topic. IMO, Safe should not try to favor anyone. Rather, it should endeavor to fairly reward each participant according to their useful contribution. Otherwise, such a well-meaning system could actually be gamed while discouraging contribution of performance to the network.

5 Likes
32

If there are elders in Paris, New York and Sydney and chunk holders in Berlin, Boston and Melbourne, which vault will be the fastest, since Paris elder would say Berlin, New York elder would say Boston and Sydney elder would say Melbourne?

To take my confusion with the idea of ‘fastest’ a step further, reliable message deliver (RMD ie rfc0058) has the idea of ‘delivery groups’ so every hop has multiple paths for every message, all paths having different latencies. The various combination of senders and receivers means every receiver sees a different order of arrival and there’s not really a true ‘first’; everyone has differing orders.

Am I misunderstanding how ‘first’ would work?

In bitcoin SAFE, hashes chunks are used for reward.

Hashes Chunks mainly depend on energy bandwidth.

Some miners farmers try to do hashing deliver chunks with less energy bandwidth cost than others (efficiency).

Therefore bitcoin SAFE rewards energy bandwidth efficiency.

The most efficient way to hash deliver chunks is to control the hashing chunk delivery with custom chips bandwidth solutions. (perhaps this is the most debatable point for SAFE?)

Custom chips bandwidth solutions get the most reward.

Bitcoin falls short of ideal reward portions because some users get more reward for the same amount of energy cost, ie they have more efficient hashing. That efficiency comes from a) better mining equipment and b) cheaper energy sources.

SAFE probably improves point A a lot and better farming equipment probably won’t give as much advantage as it would with bitcoin. But point B seems hard to avoid. Cheaper and improved bandwidth sources may end up dominating farming. It’s obviously just speculation and perhaps there’s no way or even need to address this directly, just let it be emergent from the economic properties, but it’s an interesting point of comparison I think.

Especially considering energy can be privately owned and generated and controlled (eg by solar panels) but bandwidth can’t easily be. Bandwidth is much more subject to the mercy of bandwidth providers than energy is.

It’s the difference in energy efficiency that’s the main problem, not the portionality of the rewards. The portion of the reward is based on hashing, but the benefit of the reward is based on energy efficiency, not just hashing in and of itself. So inefficient users get no benefit from their activity and that’s the true problem of bitcoin mining centralization.

The difference in efficiency is probably not something that can be coded into the farming algorithm. Might be worth thinking about how it might work though…

Not sure that I agree here. Fees are a contributing factor definitely, hopefully lightning will help this particular pain point, but the main issue is efficiency and only big players get access to a) the most efficient equipment and b) the most efficient energy sources.

The reason I’m doubtful of this is that if one person gets some reward then someone else isn’t getting it. So farming for rewards will always be competitive and people will be looking for ways to get the most reward, part of which must require reducing the reward of others, presumably by efficiency.

Reward for participation and reward for performance are not really different. High performance just looks like ‘a lot of participation’. It’s a tricky thing to negotiate.

Sorry for being pessimistic about it all. I would much prefer to be optimistic and working towards a reward nirvana (that’s why I’m here after all right!). But I feel the only way to avoid falling into similar traps as existing solutions is to face the pessimism of those traps, analyse them, find a genuine point of difference and then work towards it. So far I feel like there hasn’t really been any point of difference. Benefits from energy efficiency in bitcoin seem to directly map to benefits from bandwidth efficiency in SAFE. Maybe that’s OK in the end…?

10 Likes
33

Bitcoin falls short of ideal reward portions because some users get more reward for the same amount of energy cost, ie they have more efficient hashing. That efficiency comes from a) better mining equipment and b) cheaper energy sources.

The bitcoin system itself doesn’t care how efficient the miners are. It only cares that they produce a “winning” hash in an approximate interval, by whatever means. To my mind, an ideal system (ideally fair) would reward participants in exact (perfect) proportion to the number of valid unique hashes they produce during that interval, regardless of whether they find the winning hash or not (ala pooling). This can be calculated as block_rewards * single_miners_hashes/all_miners_hashes. If anyone disagrees with that statement/calc, please present your version of an ideally fair reward system, in bitcoin context. Once we agree on the ideal bitcoin reward distribution, productive discussion can follow, such as what could be an ideal SAFE reward distribution.

So inefficient users get no benefit from their activity and that’s the true problem of bitcoin mining centralization.

Inefficient and/or low-powered mining devices get zero benefit because bitcoin mining rewards are not sufficiently granular. This seems self-evident. In a theoretically perfect reward system, a miner who produced only a single valid hash during a block interval would receive block_rewards * 1/all_miners_hashes. That could be a very tiny number indeed, but it is never zero. The fact that they do not receive this reward, is proof that the mining reward is not granular enough and/or that the tx fees are too high to make the payment. Taken further: the fact that such a user even mining in a pool would almost certainly never receive payment is evidence again of fee barriers, but also that even the coin divisibility is not granular anough.

Now, I understand that there are all sorts of differences between miners in terms of efficiency, power costs, etc, etc. However, those are all external to the system and none of its concern. All the system should strive to do is pay out the rewards in perfect ratio to the work performed. That should be it’s sole mission, and the closer to perfection achieved, the greater decentralization will follow.

The reason I’m doubtful of this is that if one person gets some reward then someone else isn’t getting it.

Actually, that’s almost the exact opposite of what I said. Again, in an ideally fair system, everyone is rewarded in exact proportion to valid work performed.

Now, we can debate about if such an ideal system is possible/practical or not, but I think it clearly appears to be the most fair, and to strongly incentivize/promote decentralization. → See Long Tail

6 Likes
34

They get zero reward because rewards are not sufficiently granular.

The benefit is actually negative because they pay more in costs than they are rewarded. This is true even if the granularity is there.

It’s an important distinction.

I agree with your points that bitcoin would be better if it were more granular, lower fees, more decimal places, less dust etc. Bitcoin has those problems because they are needed (limited block size due to inefficient consensus mechanism leads to internal frictions like fees and dust limits). I think SAFE will probably avoid many of those problems.

Even with granularity in place and frictions removed, it’s not enough to ensure other important goals like decentralization are met.

Improving bitcoin granularity would do almost nothing to make mining less centralized because we still need small players to actually take part, which they don’t because it costs them to do that (ie they’re too inefficient).

Yes, I’m proposing something really dumb here - to internalise the external factors (such as energy efficiency etc). I’d prefer not to do that. I want to do perfect portionality and leave the messy stuff out of it. But that leads to a runaway positive feedback loop of centralization where only the most efficient can survive.

I can’t see how to get small players involved without some kind adjustment for externalities. Ignoring the external efficiency problem has caused bitcoin to become more centralized than desired. Resolving any internal roughness (like granularity and fees) doesn’t solve the bitcoin mining centralization problem.

But how to decide which externalities to account for? And in what way? That’s the point of any economic parameters!

I believe the SAFE economy is meant to create decentralization. Granularity and portionality of reward won’t achieve decentralization by itself.

Maybe I’m wrong. Maybe the economy is only meant for ensuring permanence of data and if that requires centralization then maybe that’s ok. Maybe small players should ‘have their say’ in ways other than farming. It’s definitely possible.

You say it well with “Once we agree on the ideal bitcoin reward distribution, productive discussion can follow, such as what could be an ideal SAFE reward distribution.” We can’t easily agree, hence the various forked bitcoins, and I think you’re right to say that portionality is the best we can do. But internal ideals have external consequences and they’re worth looking into too.

6 Likes
35

Couldn’t a factor of randomness be injected into the equation re who gets rewarded for serving data to the requester? I’m thinking something along the lines of:

As long as a farmer’s response is within x measures of the fastest response he will be put into a randomness queue where the “winner” will be selected at random from the fastest x responses and that winner will deliver the goods.

There might be a small penalty to pay in a slight delay in data delivery in this scenario but the trade-off would be a more “democratic” reward system which should act as a deterrent to domination of the payouts to “big players”.

2 Likes
36

I almost feel like a looong time ago there was something about a ‘lottery’ for farmers that were serving up the same data, similar to your randomness idea. Not sure that’s the best incentive mechanism though.

3 Likes
37

haha, I didn’t say “simple”, but simplest :stuck_out_tongue: (vs trusting client).

It’s a good point though. And I guess looking at what you’ve got there, there’s no real way to arrive at consensus on “first” without some total order mechanism of things received (which for some elders would put things out of the order they received), and has parsec has shown, would likely not be so swift. Which seems sub optimal…

Perhaps the answer there then is for each elder in each location to appropriate its own share of the rewards? That way you have seven (as things stand) potentially different rewards going out… That’s more fair from each elder’s perspective, and would (probably?) go some way toward spreading rewards among participants. Some node excelling so far and above the other nodes to respond first to several elders will get more reward. The likelihood of them being first all over the world is lower. But if they manage, they will get 7x the reward…

You could even more easily do a podium at each elder too, for further reward granularity as @neo suggest above.

5 Likes
38

I like that idea, though it would multiply bandwith by that many too, since everyone would always send to all elders just in case they’re close, depending on where we are with bandwith requirements that may be a tad of on issue.

1 Like
39

I think they just have to send to the section. And all elders (eventually) get that response. (Rather than targeting elders individually). but I’m not 100% sure there.

2 Likes
40

Not quite enough time to read all the posts here, just the first few responses and so I may have missed it, but @mav particularly…

I’m not seeing the downsides of being a pro farmer compared to an ordinary farmer being factored in. Maybe they will not be significant, but we shouldn’t assume this without demonstrating it. I’m talking mainly about costs.

The ordinary farmer has costs which largely come down to energy, and it is true that a pro farmer may have some advantage here, although I think we can say that this will be much less of an advantage than in bitcoin mining which is energy intensive.

Pro farmers need to cover their costs and not just break even, their profit must be more than they can get from investing capital elsewhere. So costs plus 5% or more likely 10%? Plus a factor to cover the risk of failure when investing - is a pretty high risk business IMO.

The costs of a pro farmer must not just be covered but exceeded in order to keep them in the game. By contrast, amateurs will be willing to stay based on much smaller levels of profit because their motivations are not purely financial. Social and psychological factors will both attract then and keep them. Factors such as inertia, altruism, belonging, hope and wild optimism keep them ‘playing’. So the amateur will need something close to 0% compared to the 5-10% (or whatever is needed to incentivise pro farmers). This alone is significant, but we haven’t even got to costs yet!

The costs include investment short and long term (developing custom solutions, finding and training personnel, purchasing equipment, setting up contacts with suppliers) and running costs (continual development of better solutions, employee costs, management, premises, keeping investors happy etc)

I’ve no idea how much these costs will be, again maybe we could get figures from miners’ (accounts) as a baseline, and also to see how these vary over time, profits, estimate risk of failure etc

It is also hard to know how this affects the balance between the two groups. It seems to me that it could be significant, perhaps even crucial.

BTW, keep being the pessimist @mav. It’s normally my role, and that means I can easily appreciate it! :sunglasses:

13 Likes
41

I agree that energy differentiation should be much less of an issue with safe farming compared to bitcoin mining. But I think @mav is making the point that the scarce resource will flip from energy in bitcoin to bandwidth in safe. And considering that unlike energy generation which is more under the control of private individuals, bandwidth is scarcer and so parties like ISPs could become the dominant farmers on safe (and, speculation, could even lower the bandwidth they sell to individuals so they can’t effectively compete). Other parties with the know-how and funds to develop rigs with super efficient bandwidth relays or locate in regions with better bandwidth would also have an advantage.

Still, I continue to believe that the farming centralization tendency in safe should be much less than with blockchain mining with reward apportionment based on contribution thanks to safemoney’s properties (@neo, @VaCrunch, and @joshuef all suggested ideas towards achieving that). This is assuming that ISPs continue to fairly sell bandwidth to all private citizens equitably and that governments do not put their thumbs on the bandwidth scale either.

5 Likes