Topic: Will mining become more or less centralised? - page 2. (Read 2996 times)

I don't see why "Market Forces" would counter-act the proposed action that meow mentioned, which I will restate so that you can elaborate:

1) ASIC sellers create hardware
2) Buyers buy up ASICs
3) Difficulty reaches a point where demand begins to dry up for ASIC hardware (ROIs become less attractive)
4) ASIC sellers now sitting on money, fabrication equipment and connections, an established design specification no one else has (except by reverse engineering), and little to no demand for their product begin to produce ASICs at cost and mine with them.

Where is the profit to be made for a new company to enter the market, invest in creating and designing an SHA-256 hashing ASIC (an expensive and time consuming project), contract fabrication and construction, for sale at slim profit per unit, when demand, already at a low to begin with, decreases even further with the increasing difficulty due to existing manufacturers bringing heavy hashing power into the network? If anything "market forces" would only drive said new company to enter the ASIC busines to mass-produce ASICs for themselves, not for resale, marginally reducing centralization, but not by much.

I agree that super-centralization of hashing power will in essence, destroy bitcoin, but from a business perspective, where is the invisible force that would stop that? As you said, if there is a profit to be made, a company will attempt to make it.

As for jalapenos, how are those different from the existing casual GPU miners? We will be looking at a network of some 200-250TH in the next 2 or 3 months, if the shipping dates from the major suppliers prove correct, and could easily hit 500TH by next year. A 4.5GH/sec box would be roughly the equivalent of a single very low-power video card (like a 6770) in todays GPU network (for the 250TH scenario after the halving), still requires a computer to be up and running for power, 2 USB connections (so more than 1 and you will be daisy chaining power USB hubs), and communication to the network like a GPU, and it can be used for nothing else. If someone is not interested in running a linux box 24/7 in some out of the way place, why would this technophobe suddenly be on-board with this relatively expensive (the price of a decent GPU) mystery box that occasionally makes them a fraction of a coin (roughly 1 coin per week of 24/7 uptime)?

And again, when I talk about centralization I don't mean that no one anywhere would be mining on a small scale, but simply that the contribution from the small timers would become less and less of the whole pie. Some numbers to play with (hypotheticals):
Each jalapeno would be roughly 0.0018% of the total network @ 250TH, requiring ~555 Jalapenos for every 1% of the total hashing power.
If we were to split the network in twain, 50% Casuals on Jalapenos, and 50% "Farmers" on Singles this would require ~28,000 jalapenos, at a cost of $150 each ($4,166,666 total) and 2100 Singles, at a cost of $1299 each ($2,800,000 total). You can already see a huge centralization in this fairly generous estimation, 7% of miners control 50% of the hashing power, with only the most modest of investments (this assumes $1299 per "Farmer"). I expect that it will be in fact far more skewed than this in fact, as for every person who buys multiple jalapenos, others will be buying even more singles / minirigs.

EDIT: To add some reality to my fanciful numbers, I will use the pre-order list of confirmed orders for BFL from https://bitcointalksearch.org/topic/wait-list-bfl-sc-pre-order-information-monarch-orders-too-89685
Orders:
Jalapeno SC               209
Mini Single SC              12
Single SC                  253
Mini Rig SC                  24

TH:
Jalepeno SC           0.9405
Mini Single SC            0.36
Single SC                15.18
Mini Rig SC                  36

Minirigs make up 5% of the orders and already control 70% of the projected hashing power, by order list. Jalapenos make up 40% of the orders and less than 2% of the total hashing power. This list is clearly not comprehensive, but it shows the trend fairly clearly.

I follow your point in seeing that the already Bitcoin-rich might want to to protect their investment by mining and keeping as high a proportion of daily-mined BTC for themselves as possible to prevent them from ending up on the exchanges and devaluing BTC.  However, I agree with you in seeing this as a high risky strategy and I think therefore for every one who decides to go with this there may be another or more who will not.  I don't see this particular strategy having a great influence either way.

That's the dynamic I think is the more likely outcome, as I described above.  Although the proportion of miners to Bitcoin users may decline if we're talking about a huge number of additional users, only a small proportion are required to be interested for the actual overall number of miners to increase.

I'm not sure that 'we all know'.  I tend to agree with you but I'm also listening to the opposing argument.

You are right in that by limiting my consideration to the direct correlation with gold mining, I was failing to take into consideration that a huge number of industries are dependent on raw materials, the supply of which means businesses in that industry are vulnerable to what the competition is doing.

However, I still believe my overall point stands because there is nothing else I can think of which is vulnerable in the sense of your output being entirely dependent on the number of others in the game.  If overnight the total hashing power doubles your production halves.

That's absolutely not true. First and foremost for VCs is risk of losing their investment.  ROI comes second and the higher the risk the higher the ROI and the quicker the breakeven point needs to be for it to be considered.

I would agree if those with serious bucks had good reason when it hit 10000USBTC to believe that it was actually going to stay there and that there wasn't a serious risk of so many others coming also coming into the game that their production would diminish to virtually nothing.

VCs who know what they're doing are looking for something - whether it be a patent, usp, anything to prevent competition from spoiling the party - before they part with their dollar.  As far as I can see, with the exception of the successful monopoly of efficient mining equipment (which I'll get to) Bitcoin mining has none of these.

This is a totally different situation where location, deals (honest and corrupt) with authorities, speculative land grabs and loads of other factors closing out potential competition are what holds fort.  There is no parallel with Bitcoin mining unless, we're talking about someone patenting knockout mining technology.

This brings up some really interesting points.  First what do I mean by 'massive'?  I think I mean whatever's 'massive' to the ASIC gear investor.  I was actually thinking a few tens of thousands but as you say, that's not what counts.  What does count is more than the investor can afford to lose.  And I fear some with successful GPU farms may have a false sense of security about this.  Because whatever it was that was being earned before, nobody really knows how fast or how far the initial difficulty hike will go nor where it will level out.  I happen to think other than for those with a guarantee of getting a significant proportion of their gear very early we really don't know how profitable mining will be.

As for efficiency and overheads.  It is true that your ASICs will cost a fraction of the amount to run per Mhash.  But the point is it's not very different for the person who has just the one Mini Single SC.  Neither when looking at gear nor at power consumption does the bulk buyer have a significant advantage.  It is not in the ASIC sellers' interest to give significant discounts to bulk buyers because every one out of the door is devaluing the next.  They need to extract as much as possible per unit.

My point is that these are not any more barriers to entry for the little guy as they are to the big guy.  It's about as level a playing field as you'll find anywhere.

Maybe you and I are meaning something different by our concept of centralisation.  The two considerations I mentioned in the first post (and above) are i) the proportion of Bitcoiners who mine and ii) the total number of mining operations.  I foresee the proportion going down due to mining on anything other than specialist gear becoming impractical and due to a higher proportion of Bitcoin users only interested in using the currency because it's useful and efficient and have no interest in the 'big picture'.  But as I've also said, if the numbers of users increases enormously not many need to buy something to plug in to mine - just out of novelty - for the actual number of mining outfits to increase.

On the other hand it's proportion of hashing capacity in a smaller number of hands that you appear to be talking about.  That is of interest because if too small a number get close to the 50% mark, as has been mentioned, Bitcoin is potentially in trouble.  I would be interested if there are historic figures on this too so we can see if this phenomenon is increasing, decreasing or fairly stable.  But even then, we'll need to look anew once the initial ASIC rush is over.

My conclusion is the opposite still.
As far as I can see it what someone was doing prior to now has very little to do with what's going to happen next.  As you say, a current GPU/FPGA farmer investing $40,000 has a slight advantage in terms of capital outlay than someone coming into the game for the first time spending that amount.  However that difference will be dwarfed by the unpredictability of what the proportion of the total hashing power that $40,000 will represent and when they'll get to plug them in relative to everyone else.

We simply don't know how many people are going to be buying how much in the next 6 months or year.  These $40,000 rigs may turn out to be immensely profitable with an impressive ROI and short breakeven.  But as many have pointed out, it will won't be far off the same for someone who has just bought a Jalapeno.

Unless the next generation of ASICs comes along quickly and has a similar increase in power/efficiency or the price comes down a lot I can't see why you should need to face a 'go-big or go-home' decision unless you're counting the time you need to dedicate to maintaining your rig.  That's the only reason I can see whilst it remains profitable that you'd want to go home.  Because whatever scale you're at if isn't profitable at that scale buying more isn't going to make it more profitable.  Regardless of where you are between the biggest and smallest your profitability is not going to be far off everyone else's.



OK, so let's address the monopoly thing.  From what I understand it is unlikely the next generation of ASICs and the generation after will have anything as near as the jump we're getting now.  So if there was an incentive to mine with the gear and not sell it it is now.  They do have the advantage that they will have paid for it all already by sales and can therefore afford to have a massive farm without having to repay.  But they're still having to pay the same electricity as everyone else.  Monopolists need a way of: i) preventing others entering the field (which they're not doing); ii) weakening the competition so they can take them over or; iii) put their competition out of business.

Once they've put a huge number of the same gear as they've got (or gear not quite as efficient) out there I can't see how they're going to accomplish any of this.  And...

How on earth is this rational?  Why will the demand have dried up other than because difficulty has got so high that there's not much to be made from that gear any more so nobody wants to buy it?  In which case do you think they're going to rub their hands in glee and start mining with it themselves at that point??!

I don't think data of what has been so far, not for this discussion nor for many others relating to the future of Bitcoin in forthcoming months, can tell us much because what was until now is pretty much all being thrown into the air and we won't know until it all comes back down how the landscape lies.  It's all theorising and conjecture so as much as I might talk like I'm convinced I'm right it will come as no surprise to me if I'm totally wrong!  However, it's fun for me at least being part of the process of thinking our way through the factors involved and coming to our respective conclusions.

So thanks all

Bobitza hit the nail on the head as far as I am concerned.  I don't think the network will become significantly more or less centralized.

And if it did go too centralized? Maybe hackers would try stop that.

But clearly this is an important topic and so far there has been some good discussion. But what I would like to see is some data that supports respective arguments.

Here's one: Market forces. If there is a profit to be made, another company will emerge that will decide to build and sell ASICs to the public. This "invisible force" will work in both directions: will bring companies in the market and will also keep them out.

I believe with the arrival of ASICs, the mining will become more decentralized from the number of miners perspective. BTC mining will no longer be the exclusive domain of hardware geeks that have a basement available and can build BPU farms with linux based host PCs. Anyone can buy a jalapeno and connect it via USB to their computer. No heat, no noise, no linux.

As with the GPU farms, there will be ASIC farms that have a lot of hashing power but I wouldn't say BTC becomes more centralized because of that (note that I'm excluding the scenario that ASIC producers will cease to sell to the public and become THE mining giants that you all fear). And here's a last argument for that: should this happen, I will lose trust in BTC as a fair option, I will exchange all my BTCs for fiat and get out. And I believe more of you will think the same, driving BTC prices down and leaving the greedy ASIC manufacturers with hundred of thousands of dollars tied into hardware that has a worthless output.

Thanks all,

I'm going to take some time absorbing your points before deciding if it changes my position and how.  I'll let you know

Definitely. Moreover, they all have an incentive to say they won't go private. Sell a bunch of hardware. And then go private anyways.

You should continue this train of thought

Seriously, though, I also think that producers have a big incentive to go private, developing ever-more-powerful ASICs. But then again, perhaps they will go IPO at some point...? A publicly traded mining company developing their own technology. Would you be as concerned then?

I don't agree with your FUD, but it doesn't matter. People will get used to BitPal and BitPay and BitCorp. Then they will get used to BitPal and BitCorp. Then they will get used to just BitPal. The network will still be just as reliable as ever.

Why would the hardware seller care? They have already milked demand for everything they can get.
You are just suggesting that they will sell their BTC as they mine it. Fine, I agree with that.

Crazy.
Monopoly = death of BTC

+51% hash power means network is unreliable BTC = $0

Yes, the rational business approach is to saturate the market with ASICs and then go private after demand dries up. Yes it ruins your reputation, but the reputation will be worthless once demand dries up. Expect a few large mining operations to emeege. Other miners will exit. This process will continue until there is only one miner left, BitPal Corp.

I agree completely, and I can't really think of a good rebuttal as to why they wouldn't or even shouldn't do so. A massive pre-order sell-off of their prototype units to cover the development costs of ASIC miners, basically gives them free (or at profit) closed in-house ASIC designs, and contacts in major Fab plants that see them as good money filled customers that they are happy to continue to lend machine time to, and a suite of paid-off production equipment to create the rest of the architecture. They can then produce virtually unlimited amounts of product at-cost (which could be waaaaaaaaaay below what people are paying now) to put on the network. Why wouldn't you do that?

I can only hope that the delays between ramping up production to meet demand, and the tapering off of demand to producing machines for use by the producers is significant enough for the miners to pay off most or all of their purchase, and leave them incentive to keep mining, rather than just give up when confronted with a huge unexpected difficulty spike.

One possibility, is that with continuing interest in ASICs, improvements to the architecture can both reduce cost and increase efficiency/power, creating a continual demand for new and improved ASICs, which would delay or negate the manufacturers switch from supply to mining.

As you said, we'll see...

@bcpokey


yup.  Then there's the risk of mining gear vendors going private once sales decline substantially.  They own the tech, so it would be a better profit proposition than an ordinary miner.


That's what I'm really scared of and do not believe that a 'mining gear industry' that has a single, focused market is necessarily a positive thing for miners.  At least with GPU miners do not need to worry about risk of encroachment by vendors.  But ASIC was an eventuality - we'll see where this goes.

One thing I don't like about this forum is that there are too many childboards. I remember the general tenor of a post but not enough specifics to track it down for reference. So you will have to take on faith that I'm not just making up what I say (while I will admit that I am spitballing numbers).

Anyway, I disagree. and see centralization as an inevitability. I'll start with your points and see where that leads me.

1) To say that the interconnectedness of supply will preclude a product from attracting serious investment is, well, I'll say flawed. Serious investment follows potential for serious ROI, nothing else. If USBTC skyrocketed to 10000USBTC and stayed there, you think that no one with serious bucks would attempt to capitalize on that? One might even go so far as to say that the potential for monopolizing a resource would be an attractive quality (companies love monopolies), in a high demand item. Look at diamonds, there is a glut of them, but only a very few people control the entire supply, and by dribbling out that supply, artificially inflate the price, reaping huge profits. However this is mostly irrelevant as BTC is not of the scale to worry about serious VCs (although there are supposedly VCs backing the asic makers, so...)

2) Those who will be buying "massive amounts of ASICs", being a relative term already, are likely the ones who own a "massive amount of GPUs". If you've been ticking along with a 70+GH/sec GPU farm, pulling in bitcoins, you have 2 choices: a) shutdown, throw away all your old GPUs and move on with your life. b) re-up in ASICs. [this same choice applies to hobbyists, though their response will likely be the opposite]
2.1) I would argue that choice b is more attractive precisely for the opposite reason that you suggested. Overhead costs are reduced on the order of 1000%. The fact that anyone bought an FPGA at all shows the power of overhead reduction. The $/MHash of an FPGA is vastly inferior to a GPU, extending the time for ROI. The only benefit is lower overhead. ASICs are vastly superior even to FPGAs in terms of MHash/Watt, reducing overhead that much further. Someone with the capacity to run 200GPUs, can now run (tens of) thousands of ASICs (if they could afford them) on the same infrastructure.
Meanwhile the smaller folks who were running 10GPUs or less, who used to be able to say "well if bitcoin goes to hell at least I paid off a bunch of awesome videocards to game on" can decide to go back to gaming, or buy an expensive piece of hardware that can serve no purpose in their life other than bitcoin mining. Not as attractive to them.

To add some special sauce to the points above, and get back to why I started my thread as I did...

Mining is already an increasingly centralized institution, despite all the high barriers that you mentioned (overhead, capacity, facilities, etc). There was a post analyzing the top pools and their miners, and I believe the numbers ran something along the lines of, the top 3% of miners on average held between 25% - 37% of the total hashing power. To simplify that means if there were 100 miners, and 100 units of hashing power, the top 3% each held 8Units of Hashing power on average, and the bottom 97% held .8 on average, or 10x less (using 25%). Or 20x on average for 37%.

So we already see a trend towards centralization, and a change towards asics seems like it will only hasten in this direction for all the reasons above (easier for existing large scale miners, more cost prohibitive for small-scale hobbyists). I myself being a more middle of the road type, have already been making forecasts and calculations, and I see the rolling-over of returns into increasing power as one of the only real avenues as well, and with large scale yielding better / best returns. A bit of a go-big or go-home scenario. Not that I'm there yet, but it seems to be what may have to happen.

Large scale farms IMHO, are only for those that already have huge piles of BTC. If you have large piles you want to keep the price climbing and the user base expanding to create stability over time. You need to get as many Bitcoins as possible to hoard them too keep too many from being sold off and driving the price down. I would say it is for deeply invested persons with long term interests in BTC. I would wager this is very risky  position to be in and would require very optimistic inverstors.

I think right now miners are realizing that the returns on ASICs will be about the same as GPUs not the 50x what they were doing before. I think mining will continue and those that dropped out may come back but only after things stabilize in a few months after difficulty curves flatten out.

I think the best way to prevent centralization is to promote BTC use including direct currency investment but more as and actual payment medium. If the price increase and miners are making money more people will be interested in doing so and keep mining distributed. Because as we all know it is cheaper to home brew BTC than factory farm.

I think what BTC needs is organized advertising campaigns from users and investors. (Maybe this next comment should be in another forum section.)

For example online pharmacies would do well with BTC.  Maybe we should be emailing these companies as a community (especially those that use those services) asking if they take BTC and if no why not with a link explaining what it is.
Another idea is every time you are on ebay ask the seller if they take BTC as a payment too.

I'm sure there are other ideas that could work to help promote BTC.

I'm seeing divided opinions on the boards on this one.  The argument that it will become more concentrated in fewer 'farms' seems to come from the likelihood that after the introduction of ASIC technology only those prepared to buy dedicated hardware will be mining with any chance of profitability.  The argument that it will become less concentrated seems to be that with the difference in hashrate/$ being so little regardless of how much you're buying there is not much to be gained in economies of scale.

I'm tending to go with the latter but would be interested in the factors you envisage having an influence and why you think they will push the tendency in the direction you believe it will go.

I'll present the rationale behind my own thinking.

Bitcoin mining will never attract the kind of level of serious industrial investment as something such as gold mining, apart from anything else because the amount of gold a company can extract is totally independent from the volume of gold others are extracting¹.  Buying Bitcoin mining gear for farming on the other hand is more like an auction (closer to a blind auction right now²).  This is because assuming everybody gets the gear they ordered the size of the slice of the available-mined-coin pie they end up with is only relative to what everyone else is also spending on mining gear.

So those spending massive money on ASICs i) accept it's a high risk strategy and are 'taking a punt'; ii) understand this better than I do and have reasons to be confident in their ROI calcs over quite some time; iii) have some inside knowledge or real good estimates on how much hashing power is going to come on line and when; iv) are being blinded by $ signs and may be in for a shock³!

It has been pointed out (can't remember the thread or poster sorry) that someone 'farming' in a dedicated facility has overheads that most hobbyists don't have.  If they don't have fantastic deals on energy (or build a facility with its own PVAs/wind/hydro turbines etc.) I can't see that there is a big advantage to be had from mining on a larger scale.  The bottom line is I can't see how breakeven point for a 'farmer', can be that much sooner than someone buying a hobby piece of kit.  Unlike real farming where a bigger more efficient tractor can cover vast acerage on a large farm for the same cost as a homesteader ploughs his little plot, with Bitcoin mining they really don't have an advantage.

Getting back to the question of whether it will become 'more or less centralised' this is also relative.  Because I'd guess right now (on as blunt an instrument as my impression of threads on this forum) there's a reasonable proportion of Bitcoiners who also mine.  So if a fair chunk of these switched off and didn't upgrade to ASICs whilst others go massive then it is more centralised.  However this only remains the case if the total number of Bitcoiners remains near what it is now.  If the proportion of Bitcoiners who mine dropped significantly whilst the total number of Bitcoiners increases we could still see a larger number (i.e. more distributed mining base) in absolute terms.

Thoughts?

¹ Of course the market will affect profitability but that's a separate issue.  I'm talking about a comparison of the actual 'stuff' mined in the case of gold v. Bitcoin.

² Edit:  Thinking about it, it's not like a blind auction at all!  In a blind auction you know how much you have bid and know what you are getting should you win.  You just don't how you bid in relation to others and you don't have an opportunity to re-evaluate and re-bid.  With pre-paid ASIC gear all you know is what you have paid.  How much others are buying and over what period is guesswork which means what one will receive in terms of the size of the slice pie is an unknown.

³ For this reason I really hope there aren't many who are borrowing to buy lots of ASIC kit.  Because what's happening now is a really good example of a time when one shouldn't invest more than one can afford to lose.  And by my book, borrowing heavily on such a high-risk strategy is bonkers!