Let's talk about the elephant in the room: The Storj economic model (node operator payout model)

[BrightSilence]

During the town hall presentation that just premiered on youtube @john spoke again about reworking the Storj economic model including the economic incentive model for node operators.

However, no details on a direction for these changes were provided, yet feedback was requested. Alright… I guess shooting in the dark it is then (for now).

I want to address a few things:

• Storj currently runs at a loss, subsidizing both storage and egress
• Demand for Storj DCS is growing rapidly
• What does “sustainable” mean in this context

First off, lets look at the current situation.

Storj currently charges the following for their services
$4 per TB of storage per month
$7 per TB of egress per month

Storj currently pays node operators the following
$1.50 per TB of storage per month
$20 per TB of egress per month
$10 per TB of audit/repair egress per month

Going by the median healthy pieces count (mean is unfortunately, not provided in stats, we can calculate an expansion rate of 66/29=2.28x. This leads to a cost per TB of customer data of $1.50*2.28=$3.41.
Additionally, about 5% of data needs to be repaired per month. Adding 5%*$10=$0.50 of extra costs for repair of static storage. Plus additional costs of egress by repair workers.

So, that leads to a profit for Storj labs of
Per TB storage: $4 - $3.41 - $0.50 = $0.09 (without taking account egress costs for repair workers hosted in the cloud)
Per TB egress: $7 - $20 = $-13

In short, Storj runs at a significant loss currently for both egress and storage (if you take repair egress costs of repair workers into account). And this doesn’t take into account compute costs or pesky things like having to pay salaries pay for buildings and other costs of running a business. Nor does it take into account payouts for partner programs.

Alright, let’s move on to the second point. Demand for Storj DCS is growing. That’s great news, but it also means at the current pricing model Storj Labs gets the lovely reward of running even bigger losses! Yaay!
Furthermore the wording during the town hall presentation suggested demand is growing faster that storage availability (my interpretation, this wasn’t specifically mentioned, but hinted at).
If that’s the case, Storj Labs is in a bit of a pickle here. They can’t exactly do nothing or the losses will grow and they will burn through token reserves faster and faster. But they also can’t really start paying node operators less and risk losing supply (and triggering large amounts of repair).

They also can’t start raising prices for customers as that is one of the main reasons Storj DCS is starting to see so much traction.

And for further context since last quarter 3 out of 8 tranches of long term token supply owned by Storj Labs have been unlocked. The token value has dropped and this means Storj Labs is burning through those reserves faster as well.

So what does this “sustainable” future mean then?
In my opinion there is no way this new model isn’t going to suck for node operators. We all knew this was coming ever since the massive price drops for customers were announced. But growing demand for the service and lower STORJ value have moved that inevitable moment a lot closer. I think we as node operators should start discussing what we would find reasonable. Without a starting point from Storj Labs to discuss about, I’m just going to throw out numbers that may not even work for Storj Labs (the margins for this shot in the dark are much lower than they were before the customer price drop).

Possible future payouts:
$1 per TB stored
$5 per TB egress (including repair/audit)

This would drop payouts on my largest node by 55%. Without a starting point provided by Storj Labs themselves… let’s just discuss what this (optimistic) shot in the dark would mean for our setups. Can we still afford to run our nodes, would it still be worth it?

Further considerations

• Since more supply might be needed short term, surge payouts should be more frequently used to subsidize SNO’s in times of high demand compared to supply. Furthermore, surge payouts could be used to ease transition into the new payout scheme. Starting at 200% to almost match the previous payouts. Perhaps dropping by 5% each month. The surge feature would give Storj Labs a knob to turn play with to gather supply when needed, while the minimum normal payouts give node operators a guaranteed minimum payout level to scale their setups to.
• Incentives to prevent repair should be implemented: Eg. weighted selection of nodes for upload, preferring nodes that triggered less repair in the past. Payout penalties to nodes that triggered repair in the payout month. Not paying for storage during offline time for nodes. These changes should hopefully lower the amount of repair needed, which is also why I didn’t lower the repair egress as much in the wild guess for payouts above. Also, egress is egress, it doesn’t really make sense for node operators that some egress is paid more than other.

@john: Sorry if I’m pre-empting planned discussions. But as you can imagine, this is of great concern to us node operators. Many of us have invested in setups and we need to start figuring out whether the possible future still makes sense with our setups. Please feel free to correct any (of the many) assumptions I made here.

[BrightSilence]

During the town hall presentation that just premiered on youtube @john spoke again about reworking the Storj economic model including the economic incentive model for node operators.

However, no details on a direction for these changes were provided, yet feedback was requested. Alright… I guess shooting in the dark it is then (for now).

I want to address a few things:

• Storj currently runs at a loss, subsidizing both storage and egress
• Demand for Storj DCS is growing rapidly
• What does “sustainable” mean in this context

First off, lets look at the current situation.

Storj currently charges the following for their services
$4 per TB of storage per month
$7 per TB of egress per month

Storj currently pays node operators the following
$1.50 per TB of storage per month
$20 per TB of egress per month
$10 per TB of audit/repair egress per month

Going by the median healthy pieces count (mean is unfortunately, not provided in stats, we can calculate an expansion rate of 66/29=2.28x. This leads to a cost per TB of customer data of $1.50*2.28=$3.41.
Additionally, about 5% of data needs to be repaired per month. Adding 5%*$10=$0.50 of extra costs for repair of static storage. Plus additional costs of egress by repair workers.

So, that leads to a profit for Storj labs of
Per TB storage: $4 - $3.41 - $0.50 = $0.09 (without taking account egress costs for repair workers hosted in the cloud)
Per TB egress: $7 - $20 = $-13

In short, Storj runs at a significant loss currently for both egress and storage (if you take repair egress costs of repair workers into account). And this doesn’t take into account compute costs or pesky things like having to pay salaries pay for buildings and other costs of running a business. Nor does it take into account payouts for partner programs.

Alright, let’s move on to the second point. Demand for Storj DCS is growing. That’s great news, but it also means at the current pricing model Storj Labs gets the lovely reward of running even bigger losses! Yaay!
Furthermore the wording during the town hall presentation suggested demand is growing faster that storage availability (my interpretation, this wasn’t specifically mentioned, but hinted at).
If that’s the case, Storj Labs is in a bit of a pickle here. They can’t exactly do nothing or the losses will grow and they will burn through token reserves faster and faster. But they also can’t really start paying node operators less and risk losing supply (and triggering large amounts of repair).

They also can’t start raising prices for customers as that is one of the main reasons Storj DCS is starting to see so much traction.

And for further context since last quarter 3 out of 8 tranches of long term token supply owned by Storj Labs have been unlocked. The token value has dropped and this means Storj Labs is burning through those reserves faster as well.

So what does this “sustainable” future mean then?
In my opinion there is no way this new model isn’t going to suck for node operators. We all knew this was coming ever since the massive price drops for customers were announced. But growing demand for the service and lower STORJ value have moved that inevitable moment a lot closer. I think we as node operators should start discussing what we would find reasonable. Without a starting point from Storj Labs to discuss about, I’m just going to throw out numbers that may not even work for Storj Labs (the margins for this shot in the dark are much lower than they were before the customer price drop).

Possible future payouts:
$1 per TB stored
$5 per TB egress (including repair/audit)

This would drop payouts on my largest node by 55%. Without a starting point provided by Storj Labs themselves… let’s just discuss what this (optimistic) shot in the dark would mean for our setups. Can we still afford to run our nodes, would it still be worth it?

Further considerations

• Since more supply might be needed short term, surge payouts should be more frequently used to subsidize SNO’s in times of high demand compared to supply. Furthermore, surge payouts could be used to ease transition into the new payout scheme. Starting at 200% to almost match the previous payouts. Perhaps dropping by 5% each month. The surge feature would give Storj Labs a knob to turn play with to gather supply when needed, while the minimum normal payouts give node operators a guaranteed minimum payout level to scale their setups to.
• Incentives to prevent repair should be implemented: Eg. weighted selection of nodes for upload, preferring nodes that triggered less repair in the past. Payout penalties to nodes that triggered repair in the payout month. Not paying for storage during offline time for nodes. These changes should hopefully lower the amount of repair needed, which is also why I didn’t lower the repair egress as much in the wild guess for payouts above. Also, egress is egress, it doesn’t really make sense for node operators that some egress is paid more than other.

@john: Sorry if I’m pre-empting planned discussions. But as you can imagine, this is of great concern to us node operators. Many of us have invested in setups and we need to start figuring out whether the possible future still makes sense with our setups. Please feel free to correct any (of the many) assumptions I made here.

[Vadim]

main problem is not even investment but running cost, like electricity. for me it is in good month 150$, bad month 300$. for servers. i think if payouts will drop by 55% lot of people will make exit or just will kill nodes.
To by honest this big drop in price was not the best idea in my opinion, but i hope there is a plan behind that. If it would be more egress, may be it will save situation, but lot of files like cold storage.

[unrealSpeedy]

They can simply do both: Pay a Little less to SNOs, charge a Little More for customers.

[Toyoo]

@BrightSilence

Possible future payouts:
$1 per TB stored
$5 per TB egress (including repair/audit)

That would work for me, I was operating under assumption of similar long-term values. Even a bit lower.

@BrightSilence

surge payouts should be more frequently used to subsidize SNO’s in times of high demand compared to supply

I’m not sure these would help. Surge payments would essentially be unpredictable, and I don’t think operators would dedicate resources not knowing whether they’ll get surge payment or not.

What I do think would be worth thinking of is some kind of auction for fixed contracts like, a satellite owner pays a specific node for availability of a specific amount of disk space over some longer period of time. That, in turn, could be turned into similar contracts for customers in form of early deletion fee or sth like that, without changing the base storage/egrees costs.

@unrealSpeedy

They can simply do both: Pay a Little less to SNOs, charge a Little More for customers.

There might a third, temporary way: make it easier for customers to pay in Storj, raising its value. This would allow the reserves to last longer (edit: but I think Storj cannot discuss this for legal reasons).

[Vadim]

Yes Storj tokens demand will rise token price, but big demand needed

[mgonzalezm]

@BrightSilence

suggested demand is growing faster that storage availability (my interpretation, this wasn’t specifically mentioned, but hinted at).
If that’s the case, Storj Labs is in a bit of a pickle here

I don’t think that is really the case because from what I been reading here in the forum of several people wanting to drop the less profitable satellites, and my particular case with my oldest nodes that have been full for months, they are hosting TBs of data (some 50% or more of the total space in my nodes) for the test satellite [europe-north-1] that is only sitting there without any egress.

If they needed more storage why don’t they delete all that data? I have been tempted on exit all test satellites, but if they lower the payout i definitely will.

[john]

This is a great discussion and I think it’s worth clarifying that for the network to succeed, it has to be economically viable both to be a node operator and to be a satellite operator generating demand.

We’re going to be talking a lot about this topic on twitter spaces, in fireside chats, here on the forum, and if anyone in the node operator community has other communication channels they recommend, we’re open to engaging wherever it makes sense.

As always. we want this to be a transparent dialog, and so that we end up with a good balance. In the short term, we’re holding the regular twitter spaces so anyone can join and ask questions or make suggestions.

We’ll be ready to share more complete aspects of some of the proposals we have in mind as they get more fully completed. We don’t want to throw half-baked ideas over the wall and we’re getting close.

We anticipate needing more supply and over time will need to grow to exabyte scale. but in a sustainable rate.

Please join an upcoming twitter space. We’ll be sharing more specifics later this quarter with the goal of publishing the first draft of a paper Q1 for review and feedback.

In the short term, we’re moving slowly so as to not disrupt the current network. Don’t worry about sudden rash changes. There are some great thoughts in this thread. Please keep them coming.

[mihalko]

I don`t think the model has to split payments for storage and egress. I understand that it is somehow standardized payment model for big storage providers, but satellites and node operators do not have too much costs providing egress. Mainly expenses for SNOs are for purchasing and running hw.

That said, I think storj should charge more per TB stored, nothing for egress. That will probably attract more hot storage clients, but again, SNO expenses are the same for zero or big eggress.

That will balance old and new nodes, different satellites and make economic predictions easier.

[Toyoo]

@mihalko

That said, I think storj should charge more per TB stored, nothing for egress. That will probably attract more hot storage clients, but again, SNO expenses are the same for zero or big eggress.

I’m a bit sceptic regarding this statement. Egrees revenue rewards winning races, and hence providing well-connected nodes. Without it I suspect it would be economically viable to stop drives, spin them up and serve only audit requests.

[Pentium100]

I understand that the payments to node operators will have to be reduced at some point for Storj to stop losing money. However, I hope that traffic will go up enough to compensate for that. Last month my node had 1.59TB of egress, which amounts to $31.82. I’d be OK with getting $31.82 for, say, 3TB egress.

I do not know how much money small nodes make, but if someone makes $10/month and that gets dropped to $5, it may make running the node “not worth” the effort or time, even if power and hardware was free.

[Stubbsey]

I’ve got 11TB stored and i roughly make $30 a month. I’ve contemplated many times especially recently with higher energy costs whether a node is worth running, so a drop in earnings would be the deciding factor if that was to happen.

[CutieePie]

@BrightSilence

Possible future payouts:
$1 per TB stored
$5 per TB egress (including repair/audit)

Would be fine with me, if my node could opt into some sort of cold storage, where the number of deletes is much lower than now.

I kinda think as a SNO we are lucky to be paid at all - lets face it the Token value has plummeted, so the difference for small nodes between $5 per TB egress and $20 per TB won’t be noticed in the first 12-18 months, so really is not a point worth worrying about.

I think for the older larger nodes, it would be negative to see the $20 per TB drop, as there are some of the most well connected, reliable and loyal supporters of Storj and I would expect have invested money and time in scaling their hardware to support future growth, it would be bad to just pull funding from them - it is hard to tell how many nodes some of these operators run, and the impact on the network could be bad if they just switch of the nodes.

Maybe the new price model is only applied to “new” pieces - the old pieces already stored and accessed in some cases allot, remain at the old pricing model.

Also, to help adoption - I’m not a fan of pay on Egress… maybe there is option to pay on Ingress, with a window of say 60 days where ingress is free to help onboarding, then Egress is free up to the amount of data you have stored… So if you uploaded 5TB on a new account, it would be free - you would then be able to download 5TB a month, before you are charged.

Hard times ahead for Storj, will watch with interest

CP