I don’t think any technical barriers exist, though. There’s at least two very easy ways to set up a node developed by the community. This is not be a blocker to moving ahead to incentives as Oysterpack delineated as his first phase. Phase 1 is effectively done. Can’t wait for everything to be perfect. We’re ready. Let’s go.
2 Likes
This point deserves more attention. If the cap were at 2^26, each account in a pool would need to have an average stake of 13.4 million to max it out. So unless this limitation of part keys per node changes…
- Pooling will not be offered to accounts with less than millions of Algo
OR - Pooled nodes will be operating at way less than the max cap
I borrowed the “1-click node” terminology from John Woods, but “1-click nodes” understates the requirements for consensus participation. As I have discussed above, we need robust node management solutions that target different use cases:
- retail - single machine solution
- large stake pool operators - enterprise-grade node cluster solution
The retail scenarios you described are why we need robust, user-friendly node manager solutions for retail. The path to true decentralization is through retail running their individual nodes, not retail delegating their stake to large stake pool operators.
1 Like
You are mr Algorand - thank you 
1 Like
These are a couple of draft ideas:
A. Fee scaling mechanism:
The shape of the curve aims to increase the fees when Global staking % goes down, incentivizing the firing up of new nodes to capture the higher fees. Also incentivizes risk management from node operators to protect their nodes from attacks of other node operators (pool vs pool).
On the other end of the curve, once you pass the sweet-spot the curve lowers fees de-incentivizing node operators from staking, releasing the Algos into the network to be allocated into the project’s services ecosystem.
The intention of this mechanism is to strike self-regulating fee structures that create incentive vectors that switch around the inflection point of the curve.
B. A centralization mitigation/absenteeism mitigation model that incentivizes node operators (pools or individuals) to keep as much of the registered consensus Algos Online at every round.
-
The distribution of consensus incentives would be 0 for rounds with less than 80% of online validators in a round. After that threshold is crossed the rewards set for the next block get multiplied according to the curve approved by governance.
-
Three proposed multiplier shapes, linear, “S” shape, and parabolic. (Note: The “S” and parabolic shapes push the incentives away from the Network failure horizon)
PS: Please destroy these models to the ground.
1 Like
I am not a computer expert and I tried to run an algorand node using the Aust 1 click node, but it could not synced. I tried and tried, asked questions, but was directed to do this and that and this and that, i just gave up.
I then tried to run a voi node, which synced instantly. Tho I did it just to see if my hardware was the problem, it seems that my hardware could run a voi node but not an algo node.
I really tried and tried to run an algo node, but it was too complicated… even with Aust 1 click node. I tried like 2 months ago, not sure if it’s more easy now, but anyways, if any non computer expert were to fall into the same issues I had - then it’s still quite complicated to set up an algo node.
I believe
@imshivaprasad_
also developed an easy way to run a node? But I don’t think it supports windows… If it did support windows, I would have tried to run a node using his program too…
This illustrates why we need professionally supported robust node manager solutions.
Regarding the node sync up issues you faced … The voi node would sync up fast because it has very little data. To sync up Algorand mainnet nodes, it is recommended to use the fast catch up mechanism.
Does the Aust 1 click node sync the node use the fast catch up mechanism?
1 Like
yes A1CN uses fast catchup
Fair enough.
So you reckon if I buy a new desktop computer with higher power and capacity and latest chips etc, I should be able to sync it easily using Aust 1 click node?
This fast catch up you talking about is a new upgrade or something?
How do I do this. Just install Aust 1 click node again and try to sync it as usual?
The fast catchup sync mechanism has always been there.
According to @lobo, the Austin 1 click does use the fast catchup mechanism.
I haven’t used any community tool to install a local node. I installed it using the Debian distribution package on Ubuntu.
Try submitting an issue on the Github project.
P2P pooling would be executed using a smart contract paired with a managed node operator.
In that mode there would be no “minimum”, as the contract is essentially an aggregate account.
>MaxCeiling accounts will be permitted, but not paid over the threshold.
3 Likes
A lot of this sounds great, the problem is the practicality. The core team on this is single digit numbers. There is only so much money for engineering. We can do this perfectly but then it won’t release this year.
One has to balance pragmatically.
We’re shooting for, nodes that are easy enough to use, a defi based solution and a P2P pooling option.
4 Likes
Feel free to make a PR to the paper, then the technical team can review on GH.
This forum isn’t great for technical discussion, and most of the team aren’t watching it.
Thanks for this analysis.
1 Like
I would like to start by saying thank you for the amazing work your team is doing.
Rome was not built in a day, but they were laying bricks every hour. Laying a brick every day, year after year is how you build an empire. Laying those bricks is how we build Algorand into an empire.
I have a few follow-up questions:
- Will these node runner and consensus participation solutions be officially maintained and supported by the Algorand Foundation? AlgoKit is a game changer for the developer experience. I believe the same can and should be done for the consensus participation user and node operator experience.
- Are these open-sourced, or will they be open-sourced? These would be ideal to get the Algorand developer community involved and contributing.
- Can you please elaborate on the DeFi-based and P2P solutions?
Thank you my friend! We’re trying!
on 1: It will be a joint effort with Algorand Technologies. I agree, we need a seamless UX. We’ll get there.
on 2: All fully open-source.
on 3: Still being worked on, but it’s likely, alongside the usual “run a node with a partkey” we’ll also have a contract based solution where people can stake simply by sending their funds trustlessly to a contract, where they’ll earn less of course as someone needs to run the node on their behalf, and we’re aiming for one of the major DeFi venues to offer first class staking as a service, liquid staking.
2 Likes
I was thinking along approach A but we don’t want to limit how much stake can go online. The network is most secure if all algos are online improbable but possible. If we go the Bitcoin way might lead to centralization if it gets harder to mine. I think we can base the fee adjustment on throughput(more txns in block) since that’s where the fees are coming from. Higher throughput higher fees percentage so validators stay but never , low throughput less percentage but never below 20%. When the block is 100% full you get all the fees and reward
Most Node Pool Operators operate across multiple chains and are knowledgeable enough to develop their own solutions…I’m all for open-source, but if there’s a monetary incentive for node-running, companies will fund their own development of such platforms imo…did the Bitcoin Foundation fund development of ASIC miners and the software that runs them?
3 Likes
A 50m account suspended is a big deal for the network. Paying 2 algos out 50m is peanuts giving the responsibility they have. Charge .0001 of stake. This way we have no problem with funding node runners and people will be incentivized to make up for how much they lost. .0001 is 1 algos for 10k= .0001 *10k is 1 algo, 100 for 10m staked
People can game the initial going online for incentive eligibility by initially paying for small stake say 1 Algo for 10k and topping up later. That’s fine It’s not important in terms of the health of the network, should be 0 really the important thing is the suspension.
For those accounts who are suspended and want to bypass the suspension fee by moving stake to a new account or not coming back online at all you can prevent that by charging the percentage on their next transaction through fees
Some have suggested reducing rewards for suspended accounts by a percentage. Maybe you get a smaller percentage of your actual rewards. Example : If you were to get 500 algos in rewards you get 50 instead.
I don’t see the need for the minimum we already maintain a map of online accounts currently in memory it didn’t affect anything. people are going to drop off themselves because they aren’t making money and spending too much maintaining a node. As more stake is online the less chance you get, indirectly raising the minimum. Suspension will remove those nodes anyway. Although the minimums as currently suggested apply only to those who want to be rewarded I think we should reward everyone regardless of minimums. As currently suggested the game theory suggests an effective minimum for everyone because by not staking the minimum you’re getting diluted compared to those who have the minimum. why get no rewards when others are making more just by topping up.
I think you said in a tweet that min max are up for discussion leaving it up for discussion is a disaster waiting to happen. These things should be enshrined or algorithmically determined
1 Like
I don’t see the need for the minimum we already maintain a map of online accounts currently in memory it didn’t affect anything
Can you point to where in the code this map can be used to non-manually mitigate absenteeism in the critical path?
Have you considered consensus parameters? Do you understand how the distribution of stake across accounts affects the comp and data burden?
people are going to drop off themselves because they aren’t making money and spending too much maintaining a node
What is your assertion?
As more stake is online the less chance you get, indirectly raising the minimum.
No, the more stake online doesn’t change a static floor for CI, minimum changes probability for consensus execution, not for incentives yielded, which directs stake distribution.
I think you don’t understand the technical nuances of that which you are arguing for.
I think we should reward everyone regardless of minimums
OK, I disagree.
As currently suggested the game theory suggests an effective minimum for everyone because by not staking the minimum you’re getting diluted compared to those who have the minimum.
No, this doesn’t make sense.
Can you explain your posit with example?
These things should be enshrined or algorithmically determined
Enshrined by who? isn’t this the point of decentralisation?
How can you algorithmically determine an outcome predicated on a multivariate function dependent on multiple unknowns propagating in a globally distributed system?