Topic: Question on theory ... (Read 1891 times)

Bitcoin will only be deflationary once all the bitcoins have been produced. until then inflation is the rule.

One thing everyone seems to miss is that Bitcoin is deflationary. The amount of Bitcoin generated drops quickly, but as price goes up and value of smaller amounts increases, that allows new mining equipment to become profitable again at reduced Bitcoin prices.

A problem for who? It's just a market like any other; there will be winners and losers; right now the mining equipment manufacturers are big winners in terms of ROI and miners for the most part will be losers.

It is natural and normal that profit margins get squeezed to near zero over time and those who are unable to maintain a profit margin will be squeezed out of the market as they should be. Mining speculators should be knowledgeable about these risks and should be willing and able to accept them. If not, they should stay out of the market.

this is a big problem, at the end the money maked with the first purchase, are needed for the next one and so on, in this way you will never make any profit(ora at very least a tiny one...)

You're right that there's no real "law" that requires semiconductor chips to double transistor count every 18 months. Your cynical assertion that there's a collusion to limit processor speed increase between producers is unfounded. A competitive free-market can produce the same observed trends without some backdoor agreement.

Moore's law isn't a real law, it's just an implicit agreement between fabs to limit their R&D spending. For a while when AMD and Intel were really going at it CPU speeds were doubling every 12 months, rather then every 18 months. If the companies spend more on R&D refining their processes they'd all make less profit.

So interestingly, because BTC hashes are such a simple circuit compared to a CPU the R&D costs for the chip design on die shinks would be lower, you might actually see a faster decrease in feature size if bitcoin becomes big enough to justify spending more on fab R&D.

But anway, in the long run sure you'll see exponential growth along with semiconductor processes.  However, right now we're seeing hashrate growth way, way in advance improvements in semiconductor tech, like We went from 100THash to 200THash in 35 days or so.

that is the kind of exponential growth that can't continue.

So I really just meant linear on moore's "law" timescales.

Thanks everyone I sold my two miners on ebay.  I will also need to sell my Anker hub, raspberry pi kit, and usb fan.

To help your math skills:

(1 + 1/2 + 1/4 + 1/8 + 1/16 + ...) / (1 + 1 + 1 + 1 + 1 + ...) = bad

But that only works if the difficulty does not rise, or rises very slowly. Let's suppose that difficulty rises by twenty percent every two weeks. (Right now, it is rising faster than that and there seems to be no prospect of a fall below that rate at any time soon.)

So, in this case, after two weeks you have mined enough coin to buy another miner and you imagine that it will take you you less time to mine enough coin for your next miner. How wrong can you be? Although you have 5% more miners, difficulty is up by 20% so you are mining at at rate that is 15% slower than the last time. Thus, after the next two weeks, you do not have enough coin to buy your next miner. You are only 85% of the way there and you are hit with another 20% hike in difficulty so you are now mining at 68% of your original rate despite the fact that we are only four weeks into the scheme. It will take you two and a half weeks to mine enough coin for your second miner. By the end of the third two week period, you are only half way to earning enough for your third miner but you will have made enough to buy it by the end of the fourth fortnight. So, while your first miner too you two weeks to earn, your second took two and a half weeks and the third took three and a half weeks.

So by this point, eight weeks into your process, difficulty rises have reduced your mining rate to just forty percent of the original rate. You do have three more miners but that just means that you are now mining a 46% of the original rate. Over the next two weeks, you will have earned 46% of the cost of your fourth miner. Over the next two weeks, you earn another 37% of the price of a miner and you have 83% of the price of your fourth miner. You are now earning at 29% of your original rate but it only takes you another week and a half and you have earned your fourth miner. It took you five and a half weeks to earn your fourth miner. Do the maths and you can see that it will take you sixteen weeks more to earn the price of your fifth miner, you're now over six months in and you will not earn your sixth miner until long after the cost of electricity exceeds the income from mining.



In reality, it is worse that that because we did not take into account:

* The cost of electricity

* The cost of hubs, cables etc.

* The fact that you will not be mining 100% of the time. You rig will crash while you are not there to fix it or your ISP will go down for half a day.

If your original miner purchase isn't profitable than no future miner purchases will be profitable.  Buying unprofitable miners with the income from other unprofitable miners will just guarantee that your ROI is 0%.

The limit isn't wafers but profitability.

Today we don't know what future difficulty is so while we can estimate it is going to be high, very very very (probably way beyond what you imagine as worst case scenario) high, it still means some people can delude themselves into thinking THEIR purchase will arrive before difficulty goes up and they will makes 2000% ROI in a couple months.  Some will but obviously most (the overwhelming most) won't however it leads to the "I AM SPECIAL" thinking.  "Other people might lose but I won't because ... Bitcoin".

It is easy for this delusion to continue UNTIL the difficulty goes up a magnitude or two.  Lets fast forward 12 months and BFL has already delivered 500+ TH/s of rigs,  ASICMiner has built out another 300 TH/s privately and sold another 500 TH/s publicly, Avalon has delivered 2 PH/s of chips and half of them have been assembled into boards and sold by a variety of companies and throw in KNC delivering 1 PH/s of rigs.  Hashing power in 12 months (under this scenario) will be more like 4 PH/s (4,000 TH/s) and difficulty will be 600M.  Note it doesn't matter if you think these exact numbers won't happen.  Draw up your own guestimates based on pre-order numbers.  Maybe it is 5 PH/s or "only" 2 PH/s the point is that the STARTING DIFFICULT for new hardware purchases will be much higher than today.

At this point profitability isn't just based on some vague guesstimates of future difficulty, it is largely based on day 0 difficulty (i.e if I bought X and it arrived today how many BTC could I produce TODAY).  At that point say ASICMiner was offering a hypothetical second gen USB Miners, 500 MH/s for "only" 0.25 BTC.  Would you buy one?  Would anyone? Well lets see.  Lets assume the buyer assumes difficulty will remain flat for the next year. 500 MH/s @ 600M difficulty = ~0.0004 BTC per day and the break even point (assuming no power costs) would be 625 days. 

So would you buy new hardware?  Even if you are buying (or pre-ordering) new hardware today because you are "special" would you still be doing so when difficulty is ALREADY 600M?  If people don't buy/deploy new hardware the hashrate will "stagnate".  Many purchases made over the last year are likely never going to break even but until we get through this "pre-order bottleneck" people will always find ways to rationalize why they will be profitable despite the evidence to the contrary.  When the cold hard math of 600M difficulty is staring them in the face suddenly the unprofitability is in black and white and people will stop buying new hardware.  

What about Moore's Law? That would suggest that, for the same number of wafers, you would get a doubling of transistors (and thus hashes) every 18 months or so.

Sure there is, there are only so many IC wafers made, so eventually rates of increase will be linear.

Uhm, not really in the foreseeable future. If difficulty levels off it's because hash rate levels off. And there's no reason why hash-rate should level off. There's easily, easily, 10 times the current network hash-rate on backorder at these ASIC places. Anyone who has already sunk the cost of purchasing a rig will run it at least until power costs exceed USD-equivalent bitcoin generation. And new ASICs are many times more efficient per GH than old GPUs, so with enough capital someone could theoretically have a mining operation of several terahashes/S in a standard home without overloading their circuits or burning the place down.

There's a hypothetical limit for difficulty in a rational market, but we're nowhere near that. If difficulty gets really really high and the gigahashes/watt efficiency doesn't improve it's possible we see a decline, but that's lightyears away. Even then, difficulty will decline only to the point that it's around zero profit. Some people at that point will still continue to mine at a slight negative profit anyways because they like the fun of mining or don't realize the cost of electricity or expect the price of bitcoin to increase in the future.

 If it appears there's a good chance an investment in mining rigs will produce a substantial profit, people will continue to pour money into it. Since rigs themselves generate more revenue, running out of money isn't an issue.

Look at it this way.

Basically when you purchase a miner you are looking to acquire BTC and make a profit (unless you're just here to secure the network) the two best options for that are buy it off an exchange or buy hardware and mine it.

When you purchase hardware that has one single purpose (to mine BTC LDO) you need to compare what it costs (in BTC) and then project how many coins the unit will generate.  If you pay 1 BTC (or the equivalent USD) for a miner and you estimate the unit will produce 1.1 BTC (or more) you have made a profitable purchase.  If your estimation has the unit only returning .99 BTC (or less) you have "lost" money on the purchase as it would have better to hold the BTC rather than use them to buy the miner as you would have at the end more BTC by holding, 1>.99 LDO.

The number of units you have does not change profitability (it just changes how much overall profit you end up making not the % per device), buying more miners with BTC generated from existing hardware similarly doesn't change the profitability of each unit.  The factors to consider are $/GH/s, delivery (more importantly the difficulty at time of delivery) and what you personally speculate difficulty rise to be.  That is the speculative part and that is the reason why so many people have differing opinions on what a good $/GH/s ratio is.

I will say in my personal opinion there is not a currently for sale USB miner that can even come close to producing as many BTC as it will cost to purchase it, but that is my own personal opinion based on my estimation of difficulty on delivery and difficulty rise expected in the coming months.

Well you can't "just print it" there is a cost involved and to be able to make smart decisions you need to make realistic assumptions about that cost.  How many BTCs would you mine todya if it costs you 1.1 BTC to mine one?


We know it isn't (or at least most isn't) because difficulty would be much higher.  BFL alone has 200TH/s + of pre-orders.  The exact amount is unknown because not everyone discloses but one can get a good ballpark idea.  If BFL was mining with 200TH/s+ difficulty right now would be at least doubled.

I say pretend because you can literally print it yourself unlike government fiats.  However I will ask you this ... what if most of that hardware is already being ran for 'burn in/testing' purposed on a rotating basis by the companies (namely BFL) selling the hardware.

Difficulty will rise until the projected profits are too little to justify new hardware.  The problem is there is a massive amount of pre-ordered hardware where people like you (but months ago) figured they only had to buy some hardware and they will be rich overnight.  Difficulty hasn't risen YET because the hardware isn't deployed but it will (no matter what) the hardware is already paid for.

So yes difficulty will eventually slow down and may even go down slightly but that equilibrium point is at least 10x if not 20x (or more) higher than where it is now.  So at a minimum your "plan" should consider the possibility that difficulty will at least double every 90 days for the next year.  Honestly that is probably a little optimistic.  Considering difficulty will stay flat though the chance of that is 0% and even doing a "plan" based on that is just foolish.  It is a good way to lose money guaranteed.


If you think Bitcoin is "pretend money" well that is just stupid. If I have 1,000 BTC I don't HAVE to spend it on mining hardware. I could echange it for USD and buy a sports car.

I'm starting to think im too math dumb for this hobby but ... There has to be a ceiling or a level off for increasing difficulty soon dont you think?  I mean bitcoin isnt that mainstream yet and there cant be that many people with stupid amounts of real money to spend.  But I guess thats part of the flaw.  Those who adpoted early dont have to spend a real dime to buy all this.  They just use pretend money to buy it.

Well the number of units is irrelivent. If 20 will return their purchase price in mined coins in 300 days then so will 20,000 or 2 or 1.

Still assumming current difficulty for 300 days is silly.  Take a look at a year long difficulty chart and pretend a year ago you had assumed that buying a bunch of FGPAs.  Oops difficulty increased 500%.

Say it is 300 mining days to break even at current difficulty and imagine you mined for 30 days and then difficulty doubles.  Ok so you mined 30 days and have 270 days days to break even except now it doubled so it is 540 days to break even.  Well that is longer but no problem.  You are able to mine 60 days before difficulty doubles again.  So 530 days - 90 = 440 *2 = 800 days to break even.  You mine another 90 days and difficulty doubles again ...

Starting to get the picture.