You just defined what a trusted node is with different words…
3 Likes
Well, I define trust with money while you say that vetting is not about the money.
anyone got any of those blackmarket nodes… 
Wait you trust people solely if they give you money… and you don’t trust people who don’t give you money. That is pretty black and white you know anything to do with money you can’t always trust right? Money is root of all evils.
When did this switch to me personally?
We’re talking about storj nodes and vetting period. You say that trusting the node is not about money, I argue that trusting the node is about money and as such vetting can be replaced with collateral.
When you said the word “I”
And the trust is about vetting and holding a certain amount. They both work together its not one way or the other. Trust takes time and trust isn’t just given. Just because I give you money shouldn’t be a reason to just trust me.
Also Id like to point out the vetting process is much faster then it used to be I remember when it took months to get out of vetting now it takes a month, I have started and ran 10 nodes so far all of which got out of vetting an I tested graceful exit on 2 different nodes. Its not so hard to wait to get your node vetted.
You’re quick to jump to conclusions. Storj is being discussed here, nothing else.
A proper collateral payment makes a good economical reason to trust a node without a month-long period of observing it first. You’re not losing anything, so there’s no reason to not trust the node.
Vetting took over a month for my first nodes and for most recent nodes it was close to 2 months on some satellites. I expect the process to become slower as the amount of data on a satellite increases. Sure, it’s not hard to wait, but you should not have to wait without a reason.
I don’t think I did everyone has there own opinion you wrote “I” so that means you believe that to be trustful, But me I say its not trustful there for im not jumping to any conclusions. Im sure many here would want to skip the line and there are some that say people shouldn’t be able to skip the line.
There’s always two sides to trust, It can be the other way around too, Say you were able to stake an amount for your node to skip the line, and lets throw a random amount out there say its 150$ to stake.
Then maybe 3 months in your hard drive fails or your node gets corrupted somehow then you come running saying Storj stole your money, Cause your node failed. Do you think your entitled to have your money back because you risked staking an amount?
It seems pretty clear to me that this deposit would be non-refundable in the event that the node fails. Basically what you would be doing with this deposit is transferring the “new node” risk from Storj to yourself. If you take on risk it’s assumed that you are willing to lose the money if the node fails. This would have to be explicitly spelled out in legalese, but I think it’s fair to say that the SNO would be informed of this and would know what they are getting into.
At first I dismissed this “pay to skip vetting” idea out of hand as ridiculous, but the more details OP gave about the idea the more I have to say I’m not opposed to it anymore, and it actually makes sense. Here’s the summary, as I understand it.
The vetting process primarily exists to establish that:
- A node is operating correctly and will serve all data that it stores, without corruption, because:
- If a node is disqualified, the repair process is expensive. Escrow is used to cover this expense, and if it doesn’t totally cover it then the Storj business effectively has to pay the remainder.
- A node is going to be part of the network for long enough that it does not cause metadata churn on Storj’s end.
- Graceful exit is “free” to the SNO but causes metadata churn on Storj’s end, which is why it’s only available to the SNO after some amount of time, presumably where they have produced enough profit for Storj that Storj is willing to pay this cost (in database scalability to process these changes).
So, vetting is primarily not about node reliability, it’s about the financial liability of a malfunctioning or vanished node: if a node malfunctions or leaves, Storj has to pay to repair the data. Vetting is only about reliability insofar as it protects Storj from financial losses due to bad nodes.
What we’re talking here isn’t about some token payment “I’ll give you $5, you mark my node vetted,” it’s simply about mitigating this risk differently.
The current model is to withhold some amount of revenue generated by the node as well as slowly test the node until we reach some threshold where we say that the node seems reliable enough to throw more data at it. If a node fails in the first month, it wasn’t holding much data, so the financial risk of paying for repair is mitigated.
This proposal is to allow the SNO to transfer the risk to themselves by effectively pre-funding the node’s escrow account. If the node is proven to be unreliable, Storj uses that escrow amount (as usual) to pay for the process of repairing data that was on that node. The SNO is out the money that they chose to risk, and the Storj business is largely unaffected. If the node is proven to be reliable, the SNO could eventually get their money back once the node passes a certain number of audits, or has been operating correctly for some number of months.
I think a fair counterargument to this proposal is that it would take time to develop and test, and the Storj development team should be focusing on other things. However, I think the proposal itself is sound, it would just need to be fleshed out with details (how much should the deposit be, when would it be returned, etc.).
Side note: The thread title “buying reputation” gives the wrong connotation and sounds sleazy. A better title would be “SNO assuming new node risk.”
I think most people here are missing the main point - the network gains nothing from this, it can only lose, not benefit. It’s a good idea, but from the network’s perspective this is just more risk, nothing else. Also a bit of unfair; people who throw money around start earning quicker than those who don’t, even if they have same realiability. There is already collateral amount used, why would you move it to the first day and get rid of vetting process if the SNO has money? It does not prove they’re realiable one bit.
Imagine being able to get a drivers licence immediately, without passing the test, if you pay more money. You purchase a car, drive off and crash. No harm done, right?
I agree the idea is good but, personally, I would not implement it. We’re talking about 30 days, not 300.
5 Likes
i would agree along the lines of @Storgeez, the idea is okay…
but one can only show commitment not reliability… tho maybe it could be applied to multi node SNO’s
and yet there would be a similar issue, with new locations, internet connections, network and server configurations / setups that would then be in the sense of the word unvetted and thus their reliability could be in question.
not sure this is viable, but it would certainly help SNO’s gain more traction quicker, at the expense of possible network reliability… so as usual it becomes a risk calculation for Storj Labs.
I like the idea tho, but i’m not sure the math would work out in it’s favor, ofc it might be worth doing some sort of limited testing of the concept at one point… if technically possible,
the risk calculation might be near impossible to do, since it would have to predict the behavior/reliability of SNO’s and their setups…
ofc in that case it might be best just to consider that nobody knows what they are doing and vet all,
which is most likely why they settled on the current approach in the first place.
simple and easy to manage.
There is another reason - if too many nodes fail one after another, some of the customer data could be lost before the satellite can repair it.
That would be a bigger problem than the $2 or whatever that a two month old node would have in escrow.
2 Likes
This is exactly the reason for vetting. A new node must first proove its reliability otherwise data might be at risk. So if you can pay $5 to get vetted, it’s not enough collateral to keep people from doing so that might turn their back on the network a few weeks later.
The network will be more reliable if every node goes through the vetting period. It’s not about financial compensation because money is being held back for more than just the vetting period. The held back amount during vetting is really low because the earnings are really low.
tl;dr:
Removing the vetting is a risk to data reliability on the network. The financial side is unimportant for the vetting period.
2 Likes
oh i was kinda thinking along the lines of people paying the whole held back amount into escrow when making a new node… something that in my case would be close to 10$ pr TB of node
i mean isn’t the whole long term held back amount essentially the “vetting” process of new nodes…
sure we call the first month vetting, but really this verification process is a lot more than just that 1 month, i would at the very least consider it the 9month of paying into the held back escrow.
paying 5$ would be absurd, 10 TB node and you pay 100$ and then you can skip vetting and don’t have to pay into held back escrow and then ofc after 9 months or whatever you get the 50% returned and then at the 15 month mark one could do GE.
that would be some what sensible, from my point of view, maybe i should have read the entire discussion before jumping in lol.
and people would ofc get data like normal… so might create some even more pissed SNO’s that didn’t understand how it all worked and created 10 x 10tb nodes and had them all or most of them get DQ…
so not even sure that would be something Storj Labs would want… but i don’t think something in the regard of what i suggested is total impossible…
but thats just my view, i also doubt most would even consider using such a feature.
TB of data maybe, not node.
Escrow is there to partly compensate the repair costs. If the node has 100MB of data, then the repair costs would be really low. So, the escrow amount accumulates over time as the node accumulates data.
yeah so one would just have to pay upfront based on some avg that has been preset or is generated depending on what most nodes paid in escrow pr TB of data storage…
ofc then one would run into other issues… like what if people upped their TB storage afterwards…so it wouldn’t fit in with the current technical options built into the storagenodes.
but as a thought experiment it could be viable lol seems to fail no matter how we seem to approach it.
@everyone
well have fun, i doubt this discussion is over anytime soon…
the idea doesn’t seem viable short term… maybe in a few years
Yeah, this is exactly what I was implying. You get a bunch of disappointed SNOs or in general SNOs with too much data, they go offline, you lose a chunk of the network. Returns are low enough as it is, when you get people throwing money at the problem thinking they’d get rich fast, you get a bunch of salty SNOs after a short while.
I think development time is much better spent making sure people don’t get disqualified too easily (as they have been) if they’ve been here for months or years, than gifting someone a month or two.
1 Like
What is node reliability and how is it even proved? Node is a bunch of data on a storage medium. It’s not bound to actual hardware or internet connection. So the only thing vetting proves is that someone was actually able to host a node for whatever amount of time needed for getting 100 audits.
Here’s an example: let’s say storj sees a sudden demand for data and ingress is high. As a SNO you want to get some of that data and associated profit, but your drives are full. So you go and buy some drives, but they won’t start filling until vetting is done, which means you’re missing out on that sudden demand. That’s when you put down the collateral that serves as insurance. If you set up your nodes properly, you get some data you could turn into extra profit. And if your new nodes were to fail, you lose money, storj uses money to rebuild, SNOs get your data and some repair egress. Everyone is happy and you only have yourself to blame if something happens.
Another option would be to grow a bunch of nodes on a smaller drive and deploy them when needed. In this case storj gets no collateral and potentially loses held amount the nodes would accumulate if they had more space to begin with.
actually, why do we need it to vet so fast?
as an existing SNO, you would be spinning up new node before the old nodes gets 100% filled. running them on the same IP, you get the same amount of data, vetted or not, and some of those gets directed into your vetting node… so i don’t see why you’re losing out on sudden storage demand if you are doing that?
i misunderstand something somewhere on how traffic is sent?
And if the increase in demand is so fast that existing nodes can’t handle it, I’m pretty sure, Storj would accelerate the vetting process.