Estimating early Safe Network size

Looking at the burstcoin estimated network size chart (burstcoin is a proof-of-storage blockchain using locally generated+stored data to mine burstcoins).

Burstcoin started on 2014-08-12 with 46 TB

After the first month on 2014-09-12 estimated network size was at 3618 TB

This is an average rate of 10 Gbps of data being added to the network (note this is locally generated data so did not travel over the wire, and is for the total network so is probably a much lower data rate per node depending how many miners there are). Perhaps a more meaningful figure is 115 TB per day being added.

After the first six months estimated network size was 11288 TB

This is an average rate of 5.6 Gbps of data being added to the network, or 60 TB per day.

To date, just over six years later, the size of the network is 546030 TB

Thatā€™s an average rate from the start of the network till now of 22 Gbps, or 241 TB per day, every day since the beginning of the network.

Sorta gives an interesting perspective on safe early network data rates and how much spare drive space is out there waiting to be utilized. Pretty incredible!

Do you think we can achieve these levels of growth? Is the comparison even worth anything since theyā€™re fairly different ideas?

Do you have an ideas on what the early network size might be? I think we can exceed burstcoin on all periods, I guess I would hope for at least 10x the size over the same 1/6/72 month periods? I feel probably the main limiting factor to growth will be total nodes, since each node has a pretty firm upper limit on bandwidth, but once thatā€™s parallelized across many nodes we should be able to get very high bandwidth availability. Would be interested in other estimates on initial network size if anyoneā€™s game to have a guess.

12 Likes

I would think if recursively uploading several GBā€™s of data doesnā€™t take over a day that we could. Not sure how slow upload speeds will be but that would be the biggest limiting factor when comparing to burstcoin, which you said was local storage and not being sent over the wire?

Whatever the limit is I reckon we get close or max out :sunglasses:

Isnā€™t the growth of the network proportional to the uploaded information? Ie shouldnā€™t we look at how much people are willing to spend on networks like Sia and Storj?

Yeah, could be a good comparison, do you have any stats?

Iā€™m only watching Storj at the moment. This is the latest data posted in their forum. 8 petabytes for I think 2 years of test network and 10-12 months of officially working with clients network.

I have 4 nodes with them and judging by what they earn per month from traffic I donā€™t think that more than 10% is the real data paid by people, so maybe 800 TB is users dataā€¦

5 Likes

Interesting figures, good to have some real world data points to compare.

Odd that storj with a $60M market cap can attract 8PB but burstcoin with $6M market cap can attract 546 PB.

Is this telling us that the bandwidth aspect is 750x more important than hdd space? (storj values 1PB at $7.5M and burstcoin values 1PB at $0.01M)

The difference in value for 1PB seems very high.

Maybe itā€™s because burstcoin isnā€™t as useful as storj, so storj offers significant extra value to justify the higher price per PB?

Maybe itā€™s because storj fees are too high so they havenā€™t got much upload and therefore donā€™t need much storage? Iā€™m a little surprised the demand side of storj is only 8 PB (or 800 TB by the estimates).

Maybe storj has more speculation than burstcoin? Or a stronger economic model?

Do you have any theories about why a PB is worth so much more on storj @Dimitar ?

I think we are comparing 2 different things. In Burst, space is free for the network and it can take up unlimited space (just as Bitcoin ASIC miners are free from the networkā€™s point of view).

For Storj, from the point of view of the network, the storing is not free. They pay for the occupied space because the coins are pre-generated and there is no inflation.

Storjā€™s prices are very good for what they offer compared to the competition. They are placed as a direct competitor to Amazon S3 (object storage service), and not just as a backup option (for which there are lower prices from other companies).

I would speculate that the growth of our network will depend on how much our community is willing to actively promote the network.

Some imagine that people will easily change their habits and the network will grow on its own. I am of the opposite opinion - I think we will have to fight for every user.

It is free to upload to the network if you first launch a vault and collect some coins. But people are lazy, and at first I think the main people who will use the network are the ones who pay, and they are a small percentage.

Years of struggle await us. I hope Iā€™m wrong, because years of struggle will open us up to vampire attacks from other Safe networksā€¦

1 Like

Burstcoin just stores nonsense data, then checks that data is present as a way of proving the proof of storage.

Afaik thats how it is anyway, certainly was the case when I looked into it a few years back.

Right, but if people have 500 PB hanging around to store random data, how much of that do you reckon will come to Safe Network to store useful data? I would have thought a fair bit. Although the storj numbers suggest maybe not as much as Iā€™d imagined.

Not sure what you mean by this, the network rewards miners so it doesnā€™t seem free to meā€¦ can you rephrase it, maybe Iā€™m missing something?

I confess I donā€™t know the storj economics very well, it looks like they have a fixed price (in USD) so that limits their options for growth. If storj canā€™t afford to pay farmers then they canā€™t expand the network size. Does that sound right?

1 Like

Hopefully alot will come over, I think we would need to have greater fiat equivalent payouts for that to happen, as that seems the reason most point drives at it.

I think now some useful data may actually be stored there, from things like " cloud burst" if it ever gained traction.

1 Like

The difference is that Burst has a fixed inflation and can take on any amount of farmersā€™ growth immediately.

On the contrary, with Storj and Safe, the network is growing together in its use. Storj is what we have to compare ourselves to, because its economic model is closer to ours.

Storj is a discount token, gives a 10% discount if you use it directly. When all coins are spent, it will work on the principle of Safe with recycling (but not through an autonomous system, but through a centralized third party, which may be the company Storj or any other company that decides to use their technology).

With Safe, there is a danger that when the coins run out, farmers will move to another Safe Network. I know that according to some, the data itself is valuable and that will protect us, but I do not see why it should be valuable from the point of view of farmers.

If the farmer can make more money in another network, it is reasonable to expect that he will move his resource there. The data itself has nothing to do with the farmerā€™s actions.

There is a new phenomenon in the DeFi world - yield locust. My concern is that we may also face this phenomenon, so we need to figure out a way to reduce the effects of itā€¦

5 Likes

Hopefully ppl will also see the potential of our network, it is so much more than just the coin, the coin is just essential for it to function.
If ppl think the coin will appreciate in price, that could sway resources our way.

1 Like

People can change their habits very quickly, but not for right reasons usually. They mostly dont care if something is technologically better, doesnt steal their data and so on. Give people feeling of exclusivity and/or superiority, something they can brag about, and they will come.
We need to find the right use case to promote, not the thing we are most proud of, but something ā€œJohn from accounting and Jane from HRā€ will find cool.

Too many people running vaults (expecting to earn money) can be a problem for the network, average Joe users is what will make the network grow healhy.

5 Likes

It will all hinge on how many people are willing to pay to store, and what the cost ends up being. As a thought experiment, segment the network into provide-surplus nodes, farm-to-store-for-free, and paid users.

A farm-to-store person will probably have to provide at least the amount they want to store for free * the duplication ratio (e.g. 8) to break even over the longer term. So provide 1TB to store 125GB. This space will be eaten by duplicates though, and, taken as a whole, farm-to-store use will not net add storage capacity (your network storage impact will be about the same as your provided space)

So there are two modes of growth - the give what you take group and the pairing of surplus-paid use. I personally would easily give up 1TB + bandwidth to receive 125GB of ultrasecure cloud storageā€¦ so I can imagine that this group will be an attractive proposition to a fairly wide crowd and depend entirely on the UXā€¦ itā€™s got to be as easy as download, click install, let run on whatever machine, and voilaā€¦ ā€˜freeā€™ cloud storage.

On the margin then, space for paid users will be provided from surplus nodes, and should theoretically settle around a cost of 8x the actual storage, plus bandwidth, capital and electricity. S3 for serious capacity is about 2c or lower / GB/month. 10 TB drives are about $300, or 3c/GB ā€¦ so figure 24c/GB (and falling) + a small fraction of this for the connection and electricityā€¦ I would really expect the final storage price to be about 10 months of S3, so itā€™s good value but a hard sell at first.

4 Likes

Might be worth investigating bitcoin mining pool hopping which has been around since early 2011, sounds to me like yield locust is the same sorta thing. Do you agree or am I missing some difference between the two?

Can you clarify this. Why is this a problem?

The intention to limit resources based on how much is uploaded seems to me a little backward. If people want to contribute resources why not allow them to? Extra resources would reduce the price for upload making upload more attractive to more people. Why keep resources scarce, and what determines the ā€˜rightā€™ amount of scarcity? Why not allow 500 PB from burstcoin into the safe network? Think how cheap storage would be, itā€™d attract a lot of uploaders. Iā€™m being a little more provocative here than my true feelings, but mostly I do feel like these are important and not-very-well-answered questions. And I ask them to hopefully find answers, not to cast doubt. If people are interested in exploring the possible or ideal growth rates of Safe Network Iā€™d be very keen to hear your thoughts.

1 Like

Can you clarify this. Why is this a problem?

The tricky thing to balance here is the differing rhythyms of the feedback loops that should keep the economy balanced. E.g. I just want to store for free, but the money is trickling in (although in this case, it should be cheap to store), so Iā€™ll quit (days) - versus - the price has been decent for a while, so Iā€™ll provision a new node/more space/order a new HDD/machine (seconds to weeks) - versus - ā€œI just checked in on SAFE and noticed space is cheap so Iā€™ll upload nowā€" which could be very slowā€¦ etc.

I think that once the network reaches sufficient size these things should just become noise and allow an efficient economic convergence, but the beginning may be tough and depend a lot on believers willing to bootstrap it at any costā€¦ Ironically though, the more of these angel nodes there are, the longer it will take for farming to be economic.

There is some good design already here versus yield locustsā€¦ e.g. yields increasing as your node is promoted, meaning that there is a sunk cost to farming SAFE and making it so if you jump, it will take a while to get the same deal again. I think there needs to be some more work to support rational users, e.g. a CLI script that uploads your archive backlog in chunks at set prices.

There is also not that many other places you can goā€¦ jump between Storj, Sia, SAFE, and burstcoin? Running a node is also not terribly CPU intensive, so I imagine that it should be possible to provide to all at the same time, and adjust provision based on return.

3 Likes

You missed a vampire attack from another Safe Network. There are several options from the point of view of the economic model - vampire attack from Safe network with infinite inflation, vampire attack from Safe network with 100% pre-released coins, vampire attack from Safe network with 0% pre-released coins, etc. As we know, the network cannot shrink and if there is a temporary peak in the outflow of new nodes to another Safe network, this can destroy our network.

Yes, this is the same principle. The difference is that Bitcoinā€™s physical asik miners are very limited in what they can farm. Safe will be like DeFi - it will be super easy to switch to another Safe network to farmā€¦

1 Like

I must have missed this? When did the plan switch from merging/splitting to only splitting? Why? There must be a good reason, but it seems like a massive vulnerability.

Sections cannot merge, but does that mean network cannot shrink? Why canā€™t sections just drop away?

Edit: Or is the point that uploaded data has to be held up, and therefore it cannot be dropped away?

Thatā€™s right

1 Like