Testnet 51% possible ? today ?

[BayAreaCoins]

Which makes you wonder -- do these guys even know how to code shit? If yes, what stops them from taking testnet4? They don't want to fight evil with evil?

I do not code.  I organize.  

I/We are not attempting to be evil either.  Very much the opposite.

In fact, the majority of people behind closed doors are absolutely fine with what we are doing and are having a great time watching the show.  Bitcoin Testnet couldn't have value when Bitcoin was $10 a coin, but now Bitcoin has priced out many of the people who were able to handle $10.  Bitcoin has reached a maturity where is safe and even a good thing that Testnet has a tiny amount of value.  It is a feature at this point IMO.

My goal is to help people obtain whole testnet coins and to inspire thought, both of which I believe I am successfully executing.

Perhaps it would be worth it as a call to AltQuick.com to delist that pair.

This could be exciting!!!!  

The loss would be passed along to the customers as per the FAQ and the warnings plastered all over the trading pages.

We are attacked daily too btw.  It is no secret that weirdos want to disrupt our little free market.  Plus, exchanges are a pretty prime target for fuckery on all levels... from state actors/rogue nations to mouth breathers in their 80 yr old mothers basement drinking mouthwash.

If Testnet is worth nothing, then vote with your feet and assist it to zero.  Very simple, but most people that think it is valueless are not willing to put their money were their mouth is and exchange the "worthless" testnet for Bitcoin... it is rather strange. (Plenty do, though.  I know we have customers who are Bitcoin Mailing list participants, but none of them will say it, and I won't either.)

You would hurt holders by not holding, it seems brutally clear to me.  Sell it into the dirt and take the free Bitcoins.  

The issue is Developers with >10,000 coins are crying about faucets with no business plans running dry and people who want Testnet bad enough to spend 1 Dogecoin on it, lulzzzzzzzz.  

The good part of it is that it hurts the functionality of testnet marketplaces

This is not true.  The blockstorms *HELPED* the Testnet marketplace and users.

If you want to drown me in support tickets and make my life shitty, make it where the third worlders don't get confirmations on their transactions like in Testnet v4.  Making stuff confirm really fast is like "Thank you!"

Drying out the v3 Testnet reward also helped the market's scarcity. (Scarcity, which is already grossly assisted by Testnet being "worthless" rather than a few pennies.  See my grocery cart in the parking lot Walmart VS Aldi comparison.)

It drove up the demand for our free testnet faucet, other faucets that depended on mining to serve their customer went out of business in a week, and that led to more trading activity as well.

Testnet 3 is trading for 10x the price of Testnet 4 at the time of this post... so I'd argue that Lopp failed spectacularly and Testnet 4 shows the level of incompetence he and other current Bitcoin developers are operating at.  Testnet 4 is not being hijacked, it is being flexed on because Developers did such a poor job that it deserves to be highlighted.  I almost can't help but giggle reading the mailing list at what they think would make it more worthless.

In Bitcoin terms where Bitcoin = whatever fork is trading for more value.  Bitcoin Testnet 3 is still Testnet and v4 is a sloppy attempt at a hostile(ish) take over, which has "failed".  Perhaps failure in Bitcoin Testnet makes that Testnet the "real" one (opposite of Bitcoin, lol!  Come on, this is kinda fun...)

To think we aren't ready and excited for Testnet v5 would be silly. (Testnet5.com)

I intend on turning BitcoinTest.net into a pretty sweet website in the near future as well.




Just because someone has 51% doesn't mean they are going to attack it.  I've cross that point quietly without anyone noticing way more than once.  Cryptocurrency isn't actually about security, it's about mostly about the illusion of security (until it isn't lol ).  If you have 75% of the network, split it up into 3, so it appears three entities have 25% each... and everyone is soooo thrilled by how decentralized it is!!!!! Omg it's gods gift to man because it's so fukin decentralized.  

Today I'll probably cross 51% in Testnet for fun, just to test and see how the network reacts with this new public pool (https://testnet4.btclab.dev/#/) I found last night.  It's a little risky, but that is Testnet... get out there, do some wild shit and have fun!

Edit 4/29 I'm over 51% in Testnet. Testing will last ~24 hours. https://testnet4.btclab.dev/#/app/tb1q9qa5vcsz2nwd3etl46y56z7ckj75zjd5e8c5z3

[LoyceV]

So the cost (in missed revenue) of this 51% attack to anyone who owns this ASIC would only be $21 per day.

This got me thinking, and correct me if I'm wrong: someone with enough ASIC power could not just do a 51% attack, they could build a completely new chain for testnet 4 from the very first block! There are currently 78970 blocks, many of them are CPU blocks. If I understand "chainwork" correctly, someone could start from the first block, keep quadrupling the difficulty every 2016 blocks, and reach a high enough difficulty to have the chain with most work in it at much less than 78970 blocks.

I did some math on this
(note: not all blocks are the latest blocks, while creating this post my testnet 3 node was still syncing and that took a day)

Bitcoin mainnet:

Code:

  "blocks": 895785,
  "chainwork": "0000000000000000000000000000000000000000c199c7a94d753ebba32eed8c",

If I understand correctly how chainwork works (and correct me if I'm wrong!), this hex number converts to 59916515419931912332737703308 in decimal.
The next block:

Code:

  "blocks": 895786,
  "chainwork": "0000000000000000000000000000000000000000c19a33ffa3f384b55f48b90a",

59917027028164548466674546954 in decimal.

Subtracting those numbers gives the chainwork added by the last block: 59917027028164548466674546954 - 59916515419931912332737703308 = 511608232636133936843646.
For this amount of work, the miner earned Block 895786 block reward:

Subsidy + fees   ‎3.186 BTC $316,215

Now let's compare this to the total chainwork done in testnet since it started
Bitcoin testnet 3:

Code:

  "blocks": 4325617,
  "chainwork": "000000000000000000000000000000000000000000001692ec68803dd130a497",

106602322264487309124759 in decimal.
As fraction of the total difficulty of the last mainnet block: 106602322264487309124759/511608232636133936843646=0.2084
Replacing all of testnet 3 would take hashing power worth 0.2084 * 3.186 BTC = 0.66386 BTC ($65,889) (plus one block to create the chain with most work).

Bitcoin testnet 4:

Code:

  "blocks": 81425,
  "chainwork": "0000000000000000000000000000000000000000000002817fd0ebbb115bbc6f"

11833573071674477886575 in decimal.
As fraction of the total difficulty of the last mainnet block: 11833573071674477886575/511608232636133936843646=0.02313
Replacing all of testnet 4 would take hashing power worth 0.02313 * 3.186 BTC = 0.07369 BTC ($7314) (plus one block to create the chain with most work).

Now that I'm at it: how many times the current block difficulty would it take to replace the entire Bitcoin mainnet chain?
59917027028164548466674546954/511608232636133936843646=117,115. Multiply it with the block reward, and that mining power is currently worth 139 thousand Bitcoins. But it secures 19.86 million Bitcoins!

[stwenhao]

plus one block to create the chain with most work

Not necessarily. If someone will try to produce new chain from scratch, then the difficulty can grow in a different way, than it was growing historically. Which means, that if we had difficulty 1, and then difficulty 1.18, then an alternative chain can go right away from 1 to 4 instead, like new testnets are. And also, a new chain will start from 1, if it will reorg everything, so it will take some blocks, to jump to the current difficulty.

However, when you make some artificial chain, then you can also tweak timestamps, and pretend, that everything was mined in 2009, and there were six blocks per second. Which means, that we have 4294967295 - 1231006505 = 3063960790 seconds to use in the current consensus. Also, because the difficulty can raise four times after 2016 blocks, then if it will always reach its maximum value, then we will need only around 128 adjustments in regtest. And in mainnet, we have 32 leading zero bits, so after around (256-32)/2=112 adjustments, it will prove, that SHA-256 is broken, and will halt the chain, when chainwork will reach values around 2^256 (because then, uint256 will overflow, and the next blocks will be always rejected).

So, to halt the chain, with the highest chainwork of 2^256-1, it will take in practice 112*2016=225792 blocks, or something around that value.

[LoyceV]

So, to halt the chain, with the highest chainwork of 2^256-1, it will take in practice 112*2016=225792 blocks, or something around that value.

To reach such high difficulty, I would expect you'd still need to actually mine those blocks with very high difficulty. What am I missing here?

[stwenhao]

What am I missing here?

This change will allow you to test that edge case, even if you don't know, how to break SHA-256:

Code:

if (UintToArith256(hash) > bnTarget) //original code
if (UintToArith256(hash) > UintToArith256(params.powLimit)) //accept anything with at least difficulty=1
//or remove that condition, to accept all blocks, regardless of their Proof of Work

I ran it under regtest, and saw log2_work equal to 255, and when it went closer to 256, the network halted, and it was no longer possible to mine any new block on top of that.

Also, practically speaking, a chainwork around 2^128 would mean, that the whole network produced enough hashes, to potentially generate a single SHA-256 collision, which would be a clear signal, that we need to adjust the mining algorithm, and soft-fork some replacement (for example by requiring that SHA-256 result is below a given target, and SHA-3 of the same data meets the second difficulty).