Hacker News new | past | comments | ask | show | jobs | submit login
Mining Bitcoin with pencil and paper: 0.67 hashes per day (2014) (righto.com)
642 points by dvt on Dec 18, 2017 | hide | past | favorite | 156 comments



> It's clear I'm not going to make my fortune off manual mining, and I haven't even included the cost of all the paper and pencils I'll need.

He's not taking into account price inflation over the long term. If BTC ever cracks a million USD, he'll really regret not squeezing every last bit of hashrate he can now.


Hmm. If you can mine 15 trillion hashes per second you are likely to find one bitcoin after 486 days (derived from numbers on this site www.cryptocompare.com). The human rate of hashing is .67 hashes per day. Based on a price of $1 million for one bitcoin, that would translate to a break even salary of one million-billionth of a penny per hour. Probably be hard to find good people at that rate.


China might be a good place to setup shop?


Ignoring the racism baked into that comment, smart people in China probably have more job opportunities than ever, with greater opportunity for wealth creation. China has one of the hottest tech industries in the world: Tencent and Alibaba each have market caps above $450B USD, and Baidu, JD, and NetEase are all in the $50-80B range. All these companies are rapidly growing. Not to mention the booming startup ecosystem across the country.


Labor is cheap in China. It’s not racism to make factually accurate observations.


>It’s not racism to make factually accurate observations.

That is not as true as it sounds at first glance. For example, a statement like "blacks are more violent and less educated than whites" is factually accurate, but frames the situation in a definitely racist way if you just leave it at that because without elaboration, it makes it sound like a fact of nature.

This is the problem with "facts and statistics can't be racist". No, technically they can't be, but if your conclusion you're drawing is that certain ethnicities are just genetically more criminal, you're handling the facts in a racist (and very incorrect) way (not implying in any way that you think like that).


> That is not as true as it sounds at first glance.

No. It is exactly as true as it sounds.

The conclusions one draws from facts are whatever they are. But the facts themselves are not racist, and it is not racist to say things that are factually correct. Ever. Full stop.

Also it’s ok to be white. Saying that it’s ok to be white is not racist.


Facts themselves are not racist. The person uttering them can be.


It's creepy to say irrelevant weird things, though.


If we consider stating facts without context needed to truly understand them as being equal on level of wrongness with something like discrimination, I think we should consider exactly how many people use such poorly informed facts. Gun crime facts, wage gap facts, just about any psychology paper cited ignores the limitations of the study (which likely has very few if any replications with different populations). Even medical research if filled with poorly stated facts that drop a lot of the details needed to understand the actual study done.

Given that stating facts without full context is the norm, attributing one specific case to racism seems hard to justify.

For a red delicious apples to granny smith apples comparison, look at the treatment of, framing of, and treatment of framing of crime facts based on race verses crime stats based on sex.


Interesting discussion. So when is it in your opinion allowed to draw conclusions based on personal attributes, e.g. when is something not discriminating.


I try to stick to Wikipedia definition, because people claim racism in a lot of situation where I wouldn't consider it as such: "Racism is the belief in the superiority of one race over another, which often results in discrimination and prejudice towards people based on their race or ethnicity."

In theory, the discussed comment doesn't claim his race is superior, just stating that in China people are underpaid.

For sure he is projecting stereotypes, but that's not racism, it's a different thing.

Even the legal definition doesn't seem to make that comment fit:

"the term "racial discrimination" shall mean any distinction, exclusion, restriction, or preference based on race, colour, descent, or national or ethnic origin that has the purpose or effect of nullifying or impairing the recognition, enjoyment or exercise, on an equal footing, of human rights and fundamental freedoms in the political, economic, social, cultural or any other field of public life."


> factually accurate observations

... in this case, it's a sarcastic quip based on an exaggerated truth.


the inferred reference you're missing is the 'all asians are good at math' stereotype.


East asians are good at math. Look at rankings by country. Obviously no one would actually do this, but if you were going to, China would be a great place to set up shop.


Youbate referring to a popular worldwide math test. There's no reason to believe those test scores aren't gamed, especially in Communist China where information is highly censored in general and the known "sampled" locations where the tests are given is highly suspect


You shouldn't project your own stereotypes onto other people. China is the current king of labor-intensive low-cost mass production


Remember, a hash that takes much longer than 10 minutes is worth absolutely nothing. You don't get a coin, and you can't participate in a pool.


What if you had a time machine and a lot of paper and pencils?


Go back a few years and trade a ream of paper for the 10k pizza bitcoins.


Heck, even pay the guy 100$ and say you are taking a huge risk here.


I was thinking of mining ethereum. there was a 250$ video card that could mine 500$ worth of ethereum per year. I thought wow thats profitable, but I did some more research. the same card could mine 1500$ worth 2 months ago, and ethereum was 30% cheaper.

that would have been a losing deal to get into, and it assumed the machine was mining 24/7 without issue. I tried mining with my existing card, but it spent a substantial amount of time building the DAG, and it really wasnt worth turning my PC into a space heater.

between more people mining and hashes getting harder, the price of the coin is not as much of a factor in whether its worth mining.


If BTC cracks 1 million won't that mean there's 2.1 trillion dollars worth of USD BTC in circulation in addition to all the other currencies?


Only in the most theoretical sense. Since there is no underlying asset, you are going to run out of buyers real fast at that amount.


If the market can't clear, the price will drop until it does. Saying bitcoin has reached a certain price carries with it the implication that the market is clearing at the price, i.e. that people are indeed buying.

Million-USD transactions already occur in BTC all the time.


You can't really know that except for a few publicized ones. Those "transactions" you see are simply transfers. They "could" be a transfer to someone in exchange for USD. Or, as they are in almost all cases, they are transfers from one wallet to another both owned by the same person or exchanges that are holding large amounts of BTC moving money around internally as well.


To support your argument, according to this[0] the largest recent transaction about 3 hours ago was almost 25 million USD. 5 other transactions also exceeded 1 million USD in the last hour or so.

[0]: https://blockchain.info/largest-recent-transactions


I could be wrong, but I read somewhere that almost all of those large transactions are inter or intra exchange movements simply moving money around into different wallets. Also, they only indicate a transfer, it seems very unlikely that anyone has cashed out $25m all at once with bitcoin.


There's a VR webapp that was shared recently that shows live btc transactions. Just thought i'd share that for the 10 minutes I watched it the other day I saw numerous 10M+ transactions


Those are transfers, not necessarily transactions. No one probably traded 10m in bitcoin for USD. If an exchange transfers btc from one wallet to another for holding, it will show up on the transactions list.


You only run out of buyers if there are sellers.


I have 1 BTC but I lost the wallet. So that's one less BTC in circulation. (Literally, it is impossible for that BTC to circulate).


I'm in the same boat. CPU-mined somewhere between 1.5 and 2 BTC back in 2010 or 2011, and lost the wallet years ago. I wonder how much BTC has been lost to similar mistakes, but I guess there's no way to know or even estimate, right?


I managed to loose ~53 BTC too. I'm pretty sure a large percentage of bitcoin is gone forever.

https://www.google.no/amp/amp.timeinc.net/fortune/2017/11/25...


If bitcoin price is derived from the bid ask, most of the worlds money would have to be parked in bitcoin, does that even make sense?


"parked"? No, since no real money or asset is held to back up bitcoin. It could go to $100 million tomorrow based on a few transactions, but that wouldn't be indicative of what anyone else will be able to get for it.


Now that is just wrong. If the price went to $100m tomorrow it is because people are buying and selling bitcoin for that price.


All it takes is a single trade at a $100M valuation and, for that nanosecond, that's the value of Bitcoin.


Yes but no one really considers that to be the market price.


Yes, the exchanges would, at least until a bunch more lower transactions took place.


Money supplies are constantly inflating, every financial instrument can be looked at as a form of money, and quite a few of them are actually considered to be money, such as treasuries and traveler's cheques. According to Wikipedia, the broadest measure of money supply currently in use was 10.5 trillion, in 2013.

That cryptocurrencies are not currently considered to be money by the world economic order, it goes into a much broader range of financial assets like stocks and bonds. The S&P 500 market size reached 19.6 trillion USD in 2016

Bitcoin has a lot of room to grow before it even comes close to playing in that league. Everybody on HN should be buying crypto right now, the upside is that huge.


Except it isn't parked in any way. The money goes from the hands of the optimistic buyers into the hands of those who decided to cash out just now. The only asset "backing" Bitcoin is the people willing to buy it.


It's weird to complain that because btc is backed by no set, it it not really worth it's process in USD, which is also backed by no asset.

USD is backed by "the full faith and credit of the US government". BTC is backed by the Internet and meme magic.


21 trillions, or about 3 times the value of all gold mined in the world.


And around the market size of the S&P 500.


It would be way more effective to work a minimum wage job and buy some Bitcoin with that money.


You could squeeze much more bitcoin out of yourself if you powered asic miner with a bicycle but I can appreciate low capital costs of pencil method.


Capital costs may not be trivial if you factor in food required for brainpower.


That's an operational cost, not a capital cost. ;)

Comparing capital cost between the two is also fun. Creating a literate person who can perform arithmetic certainly requires more resources than creating a person who can "merely" pedal a bicycle. There's a lot of embodied energy in a person's education, and the cost reflects that.


Also might want to consider what is the opportunity cost of what else a literate person could do instead of mining.


Or the opportunity cost of what else could the creators of that literate person could be doing with 20 years worth of time spent on that person.


cloud mining sites are scammy though but this article narrows down the search for trustworthy cloud mining sites: https://passivetalks.com/3-trusted-cloud-mining-sites-passiv...


Fair point :)

I guess the right comparison would be how much it would cost to produce another human capable of doing the SHA-256 hash by hand rather than the "fuel" to sustain it.


That's not capital in the sense of durable assets to be depreciated. That's an operational expense.


Always faster with an abacus.


using brain is way more cost efficient


(note: the conclusion of the paper actually performs a similar calculation)

The brain uses ~300 kCal per day (roughly %20 of a human's resting metabolic rate of 1300 kCal per day) = 0.35 kWattHours per day = about 4.2 cents per day (assuming brain can be powered at average market rate of energy of 12 cents/hWattHours) = about 6.3 cents per hash (at .67 hashes/day) = about 12 sextillion days to mine one block (at current difficulty of 1,873,105,475,221) = about half a septillion dollars to mine one block.

Of course might make more sense to use the full 1300 kCal, considering that a human calculator still needs to use hands for pencil manipulation and provide power to rest of organs to stay alive. Point being this is not efficient or profitable at current difficulty.

However, assuming a pre-computer world or post-apocalyptic world without computing machines, then difficulty will be adjusted to be much easier such that it conceivably would be profitable (maybe with a slave army of human computers). Transactions and blocks could easily be sent via carrier pigeon.


I've found it works best with artisanal sharpened pencils http://www.artisanalpencilsharpening.com


Lol. From the newyorker article “I’m sure there’s somebody in India who could sharpen your pencil for $8, but if you want authentic American craftsmanship … that’s how much quality costs these days.” https://www.newyorker.com/books/page-turner/pencils-and-noth...


Well, the Paypal button actually works! I wonder if someone ever placed an order though...


There was a much lower price point originally, before he got sort of famous from the book, and subsequently a TV-show (Going Deep). The current price is basically to keep people from ordering, but still having it open if someone REALLY wants to get one.


Honestly, what is getting me is that his book on sharpening pencils is actually 224 pages long. That seems awful long on such a simple subject!


I guess the joke is that sharpening pencils is what you do when you have writer's block, so if you really have writer's block, you write a whole book on sharpening pencils?


And yet he made a half hour show with episodes such as "how to bounce a ball", "how to dig a hole, "how to open a door", "how to take a nap", etc.



That book is surprisingly nuanced, full of good advice like:

- The best paired squatting methods, for when there's no good tree available (takes a lot of trust).

- Advice to look below you before using a cliff.


Cortázar wrote some stories like "how to climb stairs."


Sounds like he must be a good talker to say the least. Charismatic or at the minimum entertaining.


It's mostly satire, but of the best variety. He spoke at my university a few years ago, he definitely knows how to entertain a crowd.

https://www.youtube.com/watch?v=KabOfnbS4TQ


It's definitely majority charisma and being funny, but a portion of the show is talking to experts in the relevant fields to discuss underlying aspects of each "simple concept" that one doesn't normally contemplate.


There's only a few sentences on each page.


Wait! The price is meant to STOP people from ordering? I naturally assumed after watching the video of his artisanal pencil sharpening techniques that the prices implied exclusivity and a cut above the rest.

I ran out to the dollar store lickety-split, purchased five dozen pencils, and put in orders for all 60 of them. Sure, it was a cool 30k, but you need your pencils sharpened properly for your entire family, maids, and baby-sitters.

He really should update his order cart for people who want to get a dozen or more sharpened at a time.

I would have ordered a full gross, but I had to run off and purchase a book for my second cousin's youngest son's friend who has a book report due. https://www.amazon.com/Trends-European-Television-Communicat...


I assume the price is set by the same logic as any other price: enough to make it worth his while to close the deal.


IIRC it was a successful business at some point ...


You can get the book for much less on amazon[1] though for a reasonable price (including a look inside). There is even a direct link to Amazon on his website.

[1]: https://www.amazon.com/How-Sharpen-Pencils-Theoretical-Contr...


The price on the web site is for the pencil, not the book.



this is my favorite pencil sharpener because it has so many other uses and the blade is easy to change: https://www.amazon.com/Stanley-10-499-Change-Retractable-Uti...


$500/pencil? Sounds similar to recent online money laundering schemes.


Nope. That link has been posted around here since about 7 years ago!

[0] https://news.ycombinator.com/item?id=2453713


I didn’t mean that I think this is anything bad, just that charging that much online for something not worth anywhere near the given price is generally questionable.


Around when this was published I looked into the feasibility of pencil-and-paper wallets. What would it take to generate a key-pair using just pencil and paper?

There are tricks like big tables of precomputed multiples of the generator that make it easier, but you still need to do hundreds of 77 digit modular multiplications. And you want to be absolutely sure you made no mistake.


It is theoretically possible, as there are just a few moving parts ( 2 tweets worth ): https://twitter.com/ribasushi/status/939954274569785344

Whether it is feasible... the amount of calculations around secp256k1 is likely practically prohibitive for pencil/paper.


Well secp256k1 private keys are just 32 random bytes (with some minor exceptions) so I think you could arrange something with a plain dice and a little bit of time.


Of course, but it would be cool if you could also calculate the public component on paper (and ideally verify it computerless as well). You could really create your bitcoin/ether/etc wallet with the public key only, whilst having an extremely low attack surface (private key never touches computer memory).


I wonder if a slightly more feasible solution of producing and verifying a solution out of simple TTL, so there's no "computer" just a application specific machine made of simple commodity parts which one could trust reasonably easily would be actually worth something.


But remember the caveats about using the same key pair for multiple transactions: https://en.bitcoin.it/wiki/Address_reuse


Generating the private key is the easy part; Just 77 rolls with a ten-sided die and re-try if it exceeds the prime. Or 256 coin flips, if you prefer binary.

It is computing the public key that is so hard.


For practical purposes I'd recommend an offline laptop, maybe running Tails OS and Electrum [0]. Public key is one thing and making transactions and signatures quite another.

[0]: https://tails.boum.org/doc/anonymous_internet/electrum/index...


A mechanical wallet would be awesome!


Now I want to read a steampunk sci-fi novel where cryptocurrency is mined on steam-powered devices. Of course, the algorithms should match the capabilities of the underlying hardware.


may as well create and sign the transaction on paper also.


For signing, both your private key and a random number k need to be kept secret. The process then involves scalar multiplication, modular inversion and two more multiplications — strictly more work than generating a key-pair.


Next: Mining bitcoin with pool of monkeys with typewriters.


Monkeys with typewriters are currently busy banging out whitepapers for ICOs.


A friend of mine writes and edits ICO whitepapers. He's very skilled and gets paid pretty well for those skills.


The fact alone that writing and editing ICO whitepapers (plural) is something an individual can do as a job makes me trust ICOs significantly less.



Oh God.

Given the quality of this gig description, at least we know that whitepapers written by that person will be immediately recognizable as scams.


And yet, people will still invest in it.

The "Useless Ethereum Token" [1] subtitles itself as so:

You're going to give some random person on the internet money, and they're going to take it and go buy stuff with it. Probably electronics, to be honest. Maybe even a big-screen television.

Seriously, don't buy these tokens.

And people still invested more than $40.000 in it [2].

[1] https://uetoken.com/

[2] https://qz.com/1023501/ethereum-ico-people-invested-thousand...


Bad example. That's one ICO I would actually buy into myself if I discovered it in time. I'm a believer in timely executed jokes, much like many other people on the Internet.


What's particularly alarming about that is he appears to have had at least three people purchase his service, two of them after the first had commented that it was substandard...

You know a market's full of suckers when even the wannabe scammers aren't discerning enough to avoid being ripped off.


Smart money has always been on selling shovels to the gold diggers.


On the bright side, maybe the cryptoworld is merely investing on better GPUs by metric-tons of cash.


Over last couple of weeks, I received a dozen or so contact requests on LinkedIn from "Blockchain and ICO experts". They don't have any other verifiable jobs on their resumes.


How is that any more objectionable than the existence of professional grant request writers?


Lawyers charge exorbitant amounts for writing contracts. Selling papers is not a new invention.


A whitepaper is not a contract. It's a description of "what does this do and why should you care", i.e. literally the thing you're buying into in an ICO.



Sure it takes skill to be a scam artist, but that doesn't mean such behavior should be tolerated.


No one is denying skills, just measuring skills on a “how many monkeys” scale.


Wow that sounds like a very morally fulfilling occupation.


This made me laugh louder then i should i guess :D


i thought they write the next Hollywood-Blockbuster?


No, they're way too good for _that_.


I think Narwals are the ones responsible for the IOTA white paper.


The actual hash computation should be mechanical. It's a straight algorithm.

The part you want to optimize is the selection of the random noise you need to make the hash come out with the appropriate number of zeros. You ask your test subjects (monkey, human, slime mold, whatever) to provide the noise, you test it with your mechanical hasher, and then either reward or punish them based on the number zeros in the hash.


"The next question is the energy cost. A cheap source of food energy is donuts at $0.23 for 200 kcalories."

We will need a BTC/Donut exchange soon.


Donut futures!


Doughnuts?



I think you could even mine Bitcoin with Turing-complete cellular automata simulated by a crowd of trained dogs, receiving movement orders by a small internet-of-things bluetooth radio (powered by IOTA?), implanted into their brains.

For what it worth...


video please



I find the article's video so interesting and easy to understand for a non-mining/technical background. I almost want to start doing it just for the sake of it (like a sudoku or mental exercice).


you will earn 0.000000000002$ a day according to cryptocompare [1]

[1] https://www.cryptocompare.com/mining/calculator/btc?HashingP...


That's still 25333000 old Zimbabwe dollars, man.


This is one of the most really hackerish thing I've seen. Bravo. Thank you for submitting.


If anyone is a math teacher of the appropriate age group, "manually calculate 2 hashes" seems like a good punishment as an alterntive to the old "write x statement y times"


I think this solves our problem: Ban hardware/software based mining. Reduce the number of 0's we're looking for. UBI is born.


Someone should ICO with a paper-only crypto currency.

The confirmations wait is gonna be a killer though.

Maybe UPS should be launching this?


Well, assuming we can somehow prevent someone from cheating and writing code to solve the problems, the confirmation wait shouldn't be any worse if people are working in pools to solve the problems and submitting found answers in an online form. If 4000 people are working for their UBI in pools the confirmations should be roughly as fast as they are today.

Again the problem would be to create challenges that "can't be computed" but can be verified with a software.


I think there is a market for selling hand-mined bitcoins as a novelty item, but selling individual hashes might be a pretty small niche.


To be able to mine at all (with whatever speed), you need to be able to complete an individual hash operation in less than 10 minutes - before the next Bitcoin block is created elsewhere on the network - because you have to base your calculation on that previous block's hash. If it takes more than a day to compute one hash, it does not work, no matter how many of them you could theoretically do in parallel.


10 minutes is the average time between the creation of the two blocks, so the time could be much greater.

Block 74638 was mined almost 7 hours after the previous one, for example, because of a serious overflowing issue.

Of course, the combinaison of "no block created in 36 hours" & "the hand-calculated block is valid" is highly unlikely, but still possible.


The network adjusted difficulty aims to average 10 minutes between blocks at all times. In order to get a point where you could feasibly calculate a hash you would have to have an extraordinary breakdown of something (like all of the hashpower vanishing) in order to get enough time before the next block.


It takes a lot of billions of hashes per bitcoin - I think your hand mined one would work out a bit too expensive for anyone to buy.


Not to mention that Bitcoins are indistinguishable from themselves. So that'd be a very niche idea.


That's not entirely true - individual addresses and/or transactions can be seen as "tainted", and those coins can be tracked through the Blockchain.

I've been a Bitcoin proponent for a long time, but this lack of basic fungibility is one of its major downsides in my opinion.


> individual addresses and/or transactions can be seen as "tainted", and those coins can be tracked through the Blockchain.

I'd say it's rather tainting transactions that use outputs of your "tainted" transaction. Personally I don't like bringing "coins" into discussion because people think it's something more material than it really is while in fact "coins" are just numbers in transaction outputs.

If you have transaction that sends 2 BTC to an address you can't taint 1 BTC out of it because in this scenario they are just numbers, like you can't distinguish both 1s in 1 + 1 = 2.

That you can track transactions back to their mining block is another matter.


You'd need a new coin. Perhaps some form of proof of work that could only be done by hand


Nike sneakers :^)


>"Currently, a successful hash must start with approximately 17 zeros, so only one out of 1.4 x 10^20 hashes will be successful."

Can someone elaborate on the math here? How do we get to 1 in 1.4 x 10^20 ?


It's probably 17 hexadecimal zeros.


Yes, as shown in the second image in the article where the successful hash starts with sixteen hexadecimal zeros.


I'm still not following can you break it down?


The hash is a cryptographically random number. For it to be successful, it must start with seventeen zeros (base 16). The chance of a random number starting like that is 1 in 16^17, which is about equal to 1 in 3 x 10^20.

I can't reproduce the calculation from the article that got 1.4 x 10^20. But it does say "approximately 17 zeros" so maybe it really meant "the first 16 hex digits must be zero and the 17th must be either zero or one". That would give close to the right numbers, if I haven't messed up.


Assuming uniformly random hashes, shouldn't it be 1/2^17?


these "broke founder" war stories are getting out of hand...


Bitcoin Paper Wallets (2015) | https://news.ycombinator.com/item?id=15302500


Articles like this remind me how little I know about bitcoin.

I just recently learnt how difficulty works, so something like this is perfect for understanding the hashing part of the protocol.


Just about every single post I’ve seen from Shirriff has been a gem. One of my favorite blogs these days.


this is great. for your next trick can you do a few transactions in the tangle iota by hand too?


Weird. I was just re-reading this post the other day when I found it in an old bookmarks..


Cool. I wonder if anyone has tried to do the transaction signing procedure manually.


You are a true geek




Consider applying for YC's W25 batch! Applications are open till Nov 12.

Guidelines | FAQ | Lists | API | Security | Legal | Apply to YC | Contact

Search: