Running multiple pools is more profitable than just a single one -> this does not lead to true decentralization!

Staking & Delegation Stake Delegation
1

Dear community,

many months ago, when I read a lot about Cardano’s design, the formulas, incentive system and how Cardano wants to accomplish true decentralization, I believed that especially the following two pool parameters will accomplish this:

  • k = sets the number of desired pools
  • a0 = motivates a single pool owner to use all his pledge for just 1 pool in order to increase the pool’s ROS

Nowadays, young Cardano mainnet is converging towards a few people controlling most of the stake.

It appears that a0 is set to the wrong value (0.3), because currently it does not motivate pool operators to dedicate all their pledge to as few pools as possible, as pooltool.io clearly shows:

pooltool_active_stake
pooltool_active_stake1056×951 130 KB

(pool clusters sorted for active stake as of September 1st, 2020)

I will pick two different pool (cluster) to illustrate the problem.

First, let’s take a closer look at the very prominent 1PCT pools, operated by 1 anonymous guy, no twitter channel, no Telegram, no nothing - just claiming highest rewards, lowest fees (see website). From all the operators, his/her pools are currently receiving a very high income (see column Epoch Pool Fees, ignore IOG, ZZZ, LEO, etc.) of about 10.300 ADA every 5 days. 1PCT actual pledge is in total 1.2 Mio ADA (for all 11 pools). So for every ADA he pledged, he receives 0,8583 % ADA (only fees!) per epoch. Per year, this accumulates to about 62,6 %. In addition, his pledge counts as stake. For that 1PCT receives an additional 5.69 % per year.

For 1PCT, pool income per year could be about 68,3% of total pledge which would be 819.480 ADA or $98.337 (with 1 ADA = $0,12). So, if the 1PCT owner keeps most of his earned ADA, his ADA will double in less than 1,5 years, allowing him to control even more pools.

Disclaimer: I picked 1PCT as example, because they control largest part of active stake (except for IOHK) of all multi-pool owner. However, I could have picked any other multi-pool owner to prove my point. For example, EMUR* is even more extreme than 1PCT when it comes to gaining pool fees vs. actual pledge. The 4 EMUR pools only have a total actual pledge of 21k ADA, but stil are offering a ROA of 5.55% to delegates and have pool income of 15.200 ADA per epch.

Now, I am comparing 1PCT to ATADA.

I chose ATADA because they have a Telegram channel, offer great support and most importantly operate only 1 pool with a very high actual pledge of 3.4 Mio ADA - which is 5th place in this “Top 30” active stake table. You will find the pool at the bottom of the list, because the list is sorted for active stake. What makes them different to the 1PCT* pools? Clearly the pool income. This pool collects 2.300 ADA as fees per epoch. So for every ADA pledged, the pool receives 0,0676% ADA as fees per epoch. Per year, this accumulates to about 4,93 %. For the staking, ATADA receives an additional 5.66 % per year.

For ATADA, pool income per year could be about 10,6% of total pledge which would be 360.340 ADA or $43.240. Clearly, it will not allow the owners to double their pool number in less than 1,5 years.

Please excuse me for this long post, but stay with me for the conclusion.

In essence that shows that income per pledged ADA is ~6 times higher for 1PCT than for ATADA. Therefore pool owners are currently clearly incentivized to spread their pledge across multiple pools. If it stays like this, than we will have 20 “1PCT” pools in only 1 year from now - and I am not kidding.

This is not what Cardano needs! Cardano prospers from distributed ownership.

What is the issue?
In my opinion parameter k is working and can be raised later, but a0 currently is not doing it’s job. One of the purposes of a0 was to increase the ROA for a pool that has a higher pledge than a lower pledged pool. This is currently not the case as this forum post confirms as well.

The impact of a0 on a pools ROA is too weak, because it makes no difference for a new delegate if he gets 5.69% or 5.66% per year. If pool A (ATADA) has a pledge that is ~30x higher than pool B (1PCT0) than this should have a substantial impact on ROA - like 1% or 2% - similar as saturation does. Even if ATADA would lower their fee from now 2% to 1% it would not cause their pools ROA to increase to make a difference.

— My clear request to the Cardano design team —
Please find an effective way to financially discourage pool owners from spreading their pledge across multiple pools.

— To all multi pool delegates —
Consider changing your delegation to properly pledged smaller pools that offer additional benefits and not just ROA. This way you will support decentralization. When delegating to a smaller pool, ROA for you will be of a similar magnitude (~5.6%) as now.

— To my fellow pool owners / community members —
I am looking forward to your thoughts / constructive feedback. It’s clear that my post will not be liked by 1PCT pool or other pool owners that run multiple pools. On the other hand I believe those operators are very clever guys, they understand the math and found a way to exploit the system, which hopefully will no longer be the case in the future.

Take care,
Tom

PS:
This is no advertising for ATADA - I merely picked them as a positive pool owner example. I am not a delegate of ATADA nor do I know the team behind ATADA personally.

21 Likes
2

Thanks for starting the discussion on this, @Tom_ADADE.

I haven’t studied enough mathematics to fully understand the equations described in the docs… but it’s quite clear to any layman that the top pools right now are indeed owned by a few operators only. It’s not hard to see that this is a vulnerability in the network. Whichever parameter is allowing this, needs to be changed.

Hopefully the core team will take notice of this.

3 Likes
3

It is more complicated, and we should come back to this in half year time or more.
Cos, gut feeling especially from a very small data set, is really not a good adviser.

2 Likes
4

a0 is insignificant now. Most of the experienced pool operators are just exploiting that fact.

People care very less about decentralization otherwise preachers of decentralization who needs to lead by example will not create multiple pools.

We also created 2 pools after seeing this enthusiasm among senior pool operators just to see what could be the impact. Obviously since we are not famous we didn’t get the early mover advantage so we will decommission the second pool soon.

PoS works in this way. More greedy, more money. Cardano has to prove their version PoS will result in decentralization but I feel it will be too late if they don’t act now as none of the protocol parameters are helpful for 1000+ pools get their operational cost back.

4 Likes
5

Thanks Tom

I think you’ve just saved me several hours of work to come up with that analysis so thank you.
So my thinking, yes the system needs changing, for me for the moment i’m being mindful of the fact we want all/everyone into ADA and then we can amend as we go, the more pools the better (ideally from various pool operators).

The fact we have a few players dominating at the moment i think is sort of a price we pay for having a successful launch of Shelley (we obviously need people running pools that have pools running well, producing blocks consistently).
On a similar point, I’m currently trying to get my head around setting up a pool (doing the training modules) and if there are people out there who would consider a JV feel free to contact me. We can do this :+1:
MachTwo

2 Likes
6

Hi Mark,

yes, Cardano certainly had a successful launch and this milestone (d=0.9) was of paramount importance. But, we as a community have now the opportunity and the responsibility to influence the further development into a direction that is best overall for the Cardano ecosystem. Time will tell if we will be able to accomplish this.

From a different view point:
Owners of big pools that are close to saturation nowadays have no other choice than to open up a 2nd or 3rd pool. Due to the current design of the system most delegates just see their pools at the top of whatever ranking and then delegate to it. When one pool reaches saturation, they then delegate to the 2nd pool of the same owner. Smaller and new pools are almost completely ignored.

The whole point of saturation was, that when a popular pool saturates, people are incentivized to delegate to other non-saturated pools of other owners. This is simply not happening. a0 was intended to prevent a single owner to divide up his pledge into many smaller pieces.

Just take a look at EMUR1 pool:
grafik

1k ADA pledge, 3.5% fee and ROA of 5.52% - this is crazy! The ROA of this pool should be like 2-3% max. I mean Yoroi is great, Emurgo is great - but this shows the problem.

Take care
Tom

5 Likes
7

One cannot reply to serveral messages at once… Same opinion here.

Let’s see how fast the 1PCT-pools of this network adjust to a higher ‘k’. Or whatever Charles’s ‘dramatic increase’ is supposed to mean…

It is really good and bad the same time, that - once you understand what to do how - it is quite easy to replicate producers nodes. Also the number of relays only has to increase slightly, they can serve multiple producers. These operations become more competitive the bigger they are. And, if one really cares about “saving” money - there is nothing peventing you from renting powerful hardware and run 10 pools on the same machine on different ports (ignoring a small downtime of all pools for kernel-updates)

3 Likes
8

True, if everything you are doing is cloud based it scales very cost efficient.

9

I remember Charles saying in one of his videos that it’s a good thing that we see IOHK staking their ADA - that it means they continue to hold them instead of dumping them into the community. And I agree that it is good to see them continue to hodl.

I also see the significant need to have reliable pool operators to keep the network up and running and prevent any noticeable downtime.

But when we have passed the 150 pools count, and with d=0.9 at the start (meaning IOG still operates most of the block-producing nodes), I think IOG is already contradicting themselves when they also operate 20 stake pools.

This is preaching the gospel of decentralization while actually acting in favor of centralization.

I think it would have been better if they just anonymously distributed their delegated stake to a bunch of small pools. This would be practicing what they preach.

I also understand their concern against favoring any particular SPO or group of SPOs. But delegating only, without openly talking about the specific pools they’ve delegated to, shouldn’t be a problem for that concern.

4 Likes
10

I believe IOHK is running 20 pools with different properties to collect data themselves and not for profit only. So, if k “dramatically” goes up like CH mentioned in his video, and IOHK still only operate 20 pools - everything is good I would say.

11

Charles has already said multiple times they will do this. But deciding who to delegate to and handling that much Ada securely takes time. Meanwhile they need to use that Ada to make money just like the rest of us.

1 Like
12

Don’t remember the details, hopefully someone who does will drop in, but I don’t believe they’re keeping the rewards.

13

It might be too late to rewind the damage. For starters, smaller pool might not survive a year. The number of pools will shrink. The highest rewards from monetary expansion are paid out earlier in the system (now), which means these pool farms will acquire even bigger stakes now. In addition ADA price is lower without smart contract, which means a single entity can accumulate a lot now than in a year. I say fix it ASAP. I complained about this problem back in the ITN days, and as early as last month

Thanks to @Tom_ADADE for highlighting the issue.

2 Likes
14

On ITN, IOHK did support pools but why would they do it in the mainnet as it is an operational overhead for them?.

I think Cardano Foundation already distributed their ADA to some guys based on their contribution. You will see a pattern of bigger delegation to some people here if you analyze it. All got same amount on the same day…find it out :grinning:

I am thinking that IOHK will opt for Coinbase custody service?.

15

Funny: Yesterday i realized, that someone tries to register 1-percent pools using letters instead of numbers for “numbering”. Until now they have 16 pools running, 1 kADA pledge, 0% margin.

16

First of all: nice analysis :+1:t2:

I understand your proposal, but I doubt whether it is possible to prevent someone (or an entity) to create multiple pools. Now it is fairly easy to recognise by the ticker/pool name, but they can use different name for their pools.

Question is: why the delegators decide to stake at these * 1PCT (or whatever) pools? Do our delegators care at all about decentralization? It doesn’t look promising based on your analysis… :thinking:

1 Like
17

Hello everybody :smiley:
This does not seem fair to me indeed. I delegate my ADAs, so I understand that me and all those who delegate rush on Emurg* replacing ADAFR for me, the next epoch and the next? :roll_eyes:
6 months is too long to change this
See u my friends.
Thanks Tom_ADADE
Regards
Vincent

1 Like
18

Tom you’ve got me checking this more than i really wanted to !

So today checked the ADApools again and found that there are now almost 250 pools giving a return of over 5% ROA, which i think for a new blockchain only having been running for less than a month is probably quite good. :smiley:

Then like Tom’s original post, i hope the team review comments here and elsewhere and make the changes to encourage even more decentralisation and more choice (but those choices also need to be made by us delegators to promote that ethos) so when your deciding on your pool take a second or so.

Have Fun
MachTwo

19

This thread, and the original post, outlines very well the discussion that I’ve also seen going on elsewhere, and the things which I’ve been considering a lot over the past week or so. I wanted to add my perspective from the point of view of someone with a very small pool, CRAB, with only ~ 363 K ADA live stake, which nevertheless was lucky enough to gain 3 blocks in Epoch 213.

Gaining those 3 blocks was very lucky, however running a sustainable stake pool shouldn’t be based on luck. Despite that, I’ve still seen much of the stake migrate away since gaining those blocks—and much of the stake remaining is in fact controlled by me, but not put into pledge. That’s mostly because from what I can see, pledge doesn’t matter, at least practically rather than theoretically. I’ve seen others comment that marketing is basically more important, whereas I came to the system in setting up a stake pool thinking the quality of service and reliability would actually be more important; I thought my extensive work with distributed and high-availability systems would be a great fit for the project, which prompted me to make the investment in time and money.

It appears I was wrong about that, but nevertheless, the fact remains that I wouldn’t trust my delegation to simply the biggest marketer, especially one with no contact channels, or equally in some cases, an operator with seemingly very little experience of mission-critical systems. I think the rewards structure of the system needs to incentivise the types of stake pool operators we want to see operating within the network, and much like the arguments put forward in recent videos regarding treasury systems for ETC, actually reward those who do the work—whether that’s development, hardware utilisation, or the extensive testing already performed by lots of people just to get everything running satisfactorily—in most cases rewarded only very minimally or not at all.

I didn’t expect to make an immediate profit in running a stake pool, but I’m not doing it purely out of love, either; it has to make good business sense, one way or the other. In my opinion, the current state of affairs make very little business sense, and has already caused me to start wondering how long I’ll be able to justify operating a stake pool for—and that’s after just one month! It’s all very well to say the parameters might be tweaked and sorted out in some months’ time, but by then many experienced operators will be long gone, leaving the vast majority of the network controlled by just a few players, at which point it will be even harder to get a viable pool delegated to than it currently is.

I feel in some way that the state of affairs in operating a stake pool from the start of the Shelley era have been missold to me, even if unintentionally; I’m not sure I would have spent the weeks of work researching (including reading all the background and mission of Cardano, running numerous quality-checks in the testnet, etc.) and developing mission-critical services within my infrastructure (including open-sourcing some of that), had I realised how things would be—both with the abysmal predicted ROIs for small pools (basically zero, in many cases), and also how discouraging it would be to operate under such terms. I didn’t even realise that ‘decentralisation’ would mean a large number of IOG pools at launch, which made getting in early even harder, and arguably set a precedent where setting up multiple pools is accepted. However, I don’t blame those who have set up multiple pools; if I were in a situation where CRAB grew that large, I would likely do so myself, in the current climate. As I understood it, one of the strengths of Cardano would be that people could act in their best financial interest, and that would be in alignment with the network’s best interests. Unfortunately, at this moment in time, this does not appear to be the case.

5 Likes
20

example / # Cardano ADA Staking with Arthur J Goldman Consultants / https://www.youtube.com/watch?v=e9eDueHLhG4

1 Like