Topic: Requesting Testnet4 tBTC - page 5. (Read 2333 times)

Isn't that because many ASIC miners are stuck on the wrong chain? Because of this:
I'm tempted to compile that version again, just to see how long that different chain is.

How about the scenario in which there are no new ASIC blocks for a day, and CPU blocks continu as intended. Would those blocks all be invalidated once the next ASIC block is created?

Well, currently, testnet3 and testnet4 are just "official pump and dump coins", as Garlo Nicon said. If you will have any kind of "persistent chain", then it will be traded.

Also, receiving new coins in a coinbase transaction, should be suspicious enough. Because they have to mature for 100 blocks in the main network, and they will disappear from the test network, long before that. Not to mention, that no ASIC miner will follow the test chain, if the same power can be used for mainnet mining, and creating a stronger chain from scratch.

So, in practice, each test chain will consist of just one weak block, on top of some real mainnet chain. It is pointless to build a longer chain, because then, you will be mining on top of some block, which is invalid by definition (because of not meeting the mainnet difficulty).

Also, hashrate-based protection can be applied to any coins, even on mainnet, without any forks. For example, you have to grind 16 bits for each coin at this address: https://mempool.space/testnet4/address/tb1qk3endeq6x0xj4pjt4zwag8wf3a629rzqr8jxd7jnmlac902wa5ysqxm9wt (which means, that you need at least 2^17 hashes, to spend both coins, 2^16 distinct hashes for each input; because sighashes force signing the current input).

If the new network will be built, with any new rules, then it could be called with "client --testnetN", and the rest could be handled by the client automatically. The "complicated" path, is what you need here and now, when nothing is built yet.

1. Any new coins will land inside a coinbase transaction, and will have only a single confirmation.
2. If any existing transaction will appear, it will be as safe, as any other zero-confirmation transaction.
3. If weak blocks will be used as a method of sharing mempools, then other ASIC miners could really pick those transactions, and make them real. The only "fake" thing is some testnet coinbase, with a single confirmation, which will always vanish. And only there, you can see any test coins. Everything else will be real, in the worst case, it can be only unconfirmed or non-standard (but always valid).

Not that difficult, as you may think. For example: all nodes, which adopted my 20 hours rule, can see each other, even if they are detached from the rest of nodes.

1. You are mining blocks with CPU difficulty.
2. The difficulty is rising, but only for ASICs, that's why there are so many CPU-mined blocks. And then, the network reaches difficulty adjustment, and consensus forces everyone to mine with ASICs difficulty, so then, CPU miners are stuck, and ASICs can pick the right chain. That's why one of the latest difficulty adjustments took 4 hours, and not 10 minutes.

Your node accepts blocks, up to 20 hours in the future. The rest of the network accepts only 2 hours in the future. Each CPU-mined block pushes the current time 20 minutes forward. After six CPU-mined blocks, the network is waiting, and rejecting your future blocks.

So, your blocks are valid, but they are not accepted yet. And then, if you would broadcast your future blocks at the same time, as other nodes, doing it "honestly", then the first-seen-safe rule is applied, and sometimes they can win, but you can sometimes win, too, if you are just faster. And if you have this re-broadcasting implemented, then welcome to network propagation games!

See block header. If the difficulty is set to 0x1d00ffff, then it is CPU-mined block, and you can challenge it with a CPU. If it is something different, then it is ASIC-mined block, and you need an ASIC, to reorg it.

And also, a single ASIC block can reorg a lot of CPU-mined blocks, and this is what developers will be doing in the future in programmatic way, through their new mining pools: https://github.com/bitcoin/bitcoin/pull/31117

I still don't understand how I am able to mine blocks if ASICs are dominating the network. Shouldn't difficulty rise exponentially?

I understand that their timestamps are future, but I don't understand how this is relevant with my blocks getting rejected by other nodes. What's the difference with their Proof-of-Work?

That's going to be very difficult to send a test-transaction to, say, another developer.

That's kinda funny

You can check the current time now. And if your node is online, then you know exactly, in your logs, when you received a given block, and what time it declared. If it differs, then you can detect it, just by reading your "debug.log" file.

Exactly. The code, shared by Garlo Nicon, with 20 hours rule (instead of 2 hours rule) is now just a tool to fork other nodes out of the official chain. You can basically trigger a fork, at any time you want, and because you accept a wider range of block times, then you can just see more forks than usual. Surprisingly, some ASIC miners didn't even implement timewarp protections, so every time, when the difficulty is adjusted, you have a chance to fork some non-upgraded ASICs, for hundreds of blocks, if your chainwork is higher, than in the BIP-94 chain.

You can do it, if you apply testnet rules, on top of mainnet. Then, you mine some mainnet blocks on CPU, you do your tests, and then, your blocks are invalidated, so there is no chance, to get any value out of them. The whole concept is called "weak blocks", because you produce a regular block, in exactly the same way, as ASICs would, but you stop, after reaching some lower difficulty (for example equal to one, if you use your CPU).

If coins will not be stale, then they will be traded anyway. Note that testnet4 is still below block number 210,000. Setting the block reward into 50 tBTC4 constantly, and disabling halvings, wouldn't change anything.

Then, you will quickly go outside 21 million coins, so you will re-introduce Value Overflow Incident (or worse bugs). Also, as in the previous idea, we are still before the first halving, and it is very likely, that test coins will be traded sooner, than after 4 years, so any halving rules are irrelevant. Also note, that there are more testnet4 coins, but testnet3 is valued more, for example because the old chain is faster, and can confirm more spammy transactions per second.

If you want to reach mainnet conditions, then you should follow mainnet mempool. It is that simple. And you can produce weaker blocks, if you want to just test, if something is valid or not.

Eight leading hex chars means 32 leading zero bits, so if you can see 8-9 zeroes at the beginning, then it is probably a CPU-mined block. But: you can always just check the difficulty, if it is 0x1d00ffff (CPU block) or not (ASIC block), and filter it by that.

ASIC miners can also mine CPU blocks. And after 20 minutes, they are forced to do so, by consensus rules.

Exactly. If after one week, you have 2016 blocks, where 1008 blocks are CPU-mined, and 1008 blocks are ASIC-mined, then the difficulty should be the same (or almost the same). But it is doubled. Which means, that consensus rules push ASICs out, until reaching 10 minutes per block on average, no matter if it is 20-minute CPU block, or any-time ASIC block. Which means, that in the corner case, CPU miners can force ASICs to mine a single block per two weeks. But: in practice, there will be more ASIC blocks than that, because CPUs push block time 20 minutes forward, instead of 10 minutes, so it leaves some room for ASICs anyway (because for example if the average block time is 15 minutes, the difficulty will be decreased).

It is compatible. Your blocks are valid. They appear in "getchaintips". They are just stale, but perfectly valid, and anyone can use "preciousblock", to pick your chain specifically, and mine on top of it.

If your blocks are not stale, then they are traded.

Of course. After 20 minutes, they are forced to do so by consensus rules. But: it is more profitable, to just keep everything 2 hours in the future. And the best way is to have a long, local chain, and share things, that are within 2 hours window.

The price per coin was much higher in the past, but CPU miners just dumped it, when they sold something around 100k tBTC4.

I don't think waiting is a good option, where the whole network can be resetted again soon. Also, testnet3 has higher price per coin than testnet4. And it will take some time, to reproduce the whole spam from testnet3 into testnet4.

Some blocks are ASIC-mined, some are CPU-mined. You can compare chainwork. Testnet3 has something around 2^75, testnet4 something around 2^70, and mainnet something around 2^95.

That's why I think all test coins should be just weak blocks, based on mainnet. But: people should try other ideas first, should see, how they fail, and then, maybe they will be convinced.

You can put any future timestamp you want inside your blocks, because sooner or later, clocks in other nodes will get there. Even if you mine mainnet coins, then still: nothing stops you from putting future timestamps. And if 51% of the mining power would set block times into 20 minutes intervals, instead of 10 minutes, then they could increase their domination for a while, and lower the network difficulty. Obviously, in that case, it will be lowered for everyone, so it will quickly increase again, when "honest" miners will use the current time. But still: putting future timestamps is profitable. Maybe not honest, but definitely profitable. In all networks, which share Bitcoin rules.

Because when you download the history, then there is no way to tell, who put the right time in the past. You can judge it, if you witness it now. But if you have a history, then you can only view it in a way, which is relative to the current timestamp. Which means, that even if there would be a strict rule, to use the current time, with higher accuracy, then, you can still connect with some new node, and share a history with fake timestamps, because Proof of Work is the only protection in that case.

Also, a little inaccuracy is needed, because there are some delays, related to propagating blocks in the network. And if you make those rules too strict, then imagine losing a block, only because your time is different by one second, than in some other node. And imagine node A, rejecting your block (one second too far in the future, sorry, rejected), and then, node B, accepting your block (because now, one second elapsed, and it is correct).

You can tell the current time here and now. But you cannot tell, if a block, mined one day ago, was honestly mined, or was it future-timestamped, or past-timestamped, for whatever reason.

Now I truly feel the 2010 vibes.  

I left my computer open for a couple of hours, and it solved blocks which are currently valuated for 50 cents each. Quite wild I'll say.

100k tBTC4 wouldn't take more than 5 days in my computer. Effectively, I would have converted a few dollars worth of electricity to $1200. I just can't believe it.

What's a "future block"? Another way to say reorg?

I can notice there is a problem. My cpuminer has mined several blocks in the last few minutes, but none of them has reached mempool.space's node: https://mempool.space/testnet4/address/tb1qh022086je57mu0klkyakdgcea5ywg300cp0ddy.


Edit: Someone sent me 0.016 tBTC4 the moment I posted. I cannot explain how this happened.

Edit #2: Someone has setup a bot that monitors this thread and funds it when a new address is posted. I really don't understand what's going on.

There's no point in replacing it again if the same thing will happen.
Ideally, this should be possible:

Or infinite inflation, or doublings instead of halvings.... But that kinda defeats the purpose: testnet is supposed to be more or less equivalent to Bitcoin, so you can test things before using real coins. All those changes are not helping, and will probably create new problems: infinite inflation for instance can lead to infinite on-chain spam and very high transaction fees, which makes it unusable again.

It was possible....

My take: the old testnet (3) market is based on scarcity and blockchain spammers (I call anything Ordinal-related spam). Does that mean the testnet (4) market is based on future scarcity expectations? Hoarding while it's cheap to sell later?

What if it's ASIC mined? Testnet shouldn't be competing with real Bitcoin for ASIC power.

Isn't this part of the problem? "Future blocks" shouldn't exist.

That answers my previous question.

Isn't the solution to make testnet much less forgiving for inaccurate timestamps? NTP (Network Time Protocol) is very accurate, and I can't remember the time I manually had to adjust a computer's time. It's always correct, and all I adjust is my own local timezone. If someone doesn't use the correct time on his system, why not just reject their blocks?
I get why the "20 minute rule" is made, it makes total sense to drop the difficulty to 1 if it takes too long to mine the next block. But this should only happen if the last block was really 20 clock-minutes ago. I have no idea what unintended consequences such a change would have.

https://altquick.com/exchange/market/BitcoinTestnet3
https://altquick.com/exchange/market/BitcoinTestnet4

And if developers will create testnet5 with the same rules, then the same thing will happen:

1. Some people will build testnet5 from the source code, and mine it.
2. This exchange will list it, when it will be officially released.
3. People will start selling their CPU-mined coins.

And if you have for example 100k tBTC4, then selling them for 0.02 BTC is really profitable, if all of that is just CPU-mined.

Exactly: a few months ago. Now, you have mining pools, which can mine future blocks, so it is much harder to compete. But of course, you can mine a lot of blocks, and see, how they vanish after a while. Or: you can trick some other nodes to follow your chain, for example when some of them didn't apply timewarp rules. Or: CPU miners can increase the difficulty to the point, where it will take ASICs a lot of time, to mine a single block per difficulty adjustment, to unstuck the network.

They are not "kept away". They are mining future blocks, and only then, they are mining with the real difficulty, when they run out of CPU blocks. Also, ASIC blocks now have future timestamps, to limit CPU-mined blocks.

But of course, if you have CPU difficulty, then you can reorg this block with just a CPU, so you have some chances. But obviously, ASIC blocks pick the chain, so they can ignore your fork, and build on top of their own chain, and then your chain will be reorged.

Where is this market?

How is this going to be any different than testnet4? The way I understand the rationale is that we reset the network create a new network if acquiring coins in the current one is only practically possible by buying it. But, it's entirely possible to mine in here, I have mined several blocks a few months ago.

What's the technical addition that keeps the ASICs away in here?

This network is slowly dying. It is traded at 20 real satoshis per 1 tBTC4, which means, that we are approaching testnet5. However, if new testnets will be started in exactly the same way as before, then it will be just an official pump and dump, signed by Bitcoin Core developers. And that means, the new testnet should have completely different rules (for example: all test coins should be stale, and disappear after a while).

Well, it works in quite unexpected way: if there is more than a single CPU miner, then ASIC miners can pick different chains. And as a result, you can fork some ASICs off the network, if your blocks will contain future timestamps.

Actually, you can mine many blocks. But: they will usually be reorged. If you have CPU block vs some other CPU block, then ASIC blocks decide, which chain is real. And also: ASICs can simply halt the chain, by not mining anything, then every CPU miner will be stuck at the next difficulty adjustment.

So that's what this guy must be doing. He mined 58 out of the last 60 blocks (and that's where I stopped counting).

More interesting (and since there's no real Testnet4 topic yet I'll post it here) is that Testnet4 is now also being spammed. See mempool.space: blocks are full, and transaction fees are >1 tBTC already. It started by creating dust, followed by consolidating dust. That address has 1.39 million UTXOs.

I don't think that works anymore. From my logs (grep 'new best'):
As you can see, any other "easy" miner gets their blocks replaced. It's as if someone has his blockchain on "read only" and ignores blocks from anyone else. That means no developer can mine a block for testing.

Exactly. There is no way to tell, if you are a CPU miner, or an ASIC miner, if both blocks have minimal difficulty. And then, you can compete.

If you have a regular signature, then it takes some bytes. If you put some Proof of Work into that, then you can make it smaller. If you check in the Script, if your signature has N bytes or less, then it requires grinding, to meet that condition. And even if all private keys are known, then still: your coins are safe, as long as nobody else will grind it faster.

More than that: you can have "signet-on-testnet", if you just put some signet challenge in your output script, and then require a signature, to contain less than N bytes. Then, signing is required, so nobody will sign those coins without permission, but also, Proof of Work will be needed, so it will be safe in a group of miners, with similar power. So, it will work more or less like Signet, but where you can peg testnet coins in and out, and where you can change signet challenge on the fly.

I guess you could mine a block or two per day, if you run your miner 24/7. As long as there is any ASIC, you can mine a block on top of that, because then, it brings us back to the past, and then, CPU mining is the easiest.

Another thing is that ASIC owners can reorg hundreds of CPU blocks, if they want. They just don't do that for some reason, but they are incentivized to do that, by testnet consensus rules.

I didn't really mean it on a technical level, I understand why the difficulty is high. The "flaw" is that other miners can make it very difficult for anyone to get testnet coins. It should be possible to get some testnet coins at any time when you need them, so there's no need to hoard them already.

I take it I did this at the right moment back then. If I'd try the same thing now, it wouldn't work.

Yes, but not all of them. I can get a block or two per day, which is still quite good result, if you consider, that I didn't improve block propagation for blocks, mined after two hours.

It is not a flaw. Quite the opposite: pushing time to the future, and CPU-mining blocks is "a flaw". But people decided to not patch it. However, high difficulty is just a consequence of testnet consensus rules. And the reset from testnet3 into testnet4 was not, because of difficulty. Actually, you can run the same tricks, to CPU-mine blocks in both testnet4 and testnet3. I also mined many CPU blocks in testnet3, so I know, that it works. But obviously, ASICs can rule both networks, as it was in the past. And in testnet4, once per 2016 blocks, it is guaranteed, that a single block will be mined by ASICs, so if you will have only CPUs in testnet4, then they will see, how the chain stops, when they encounter the nearest difficulty adjustment.

It was inevitable. And it is good, because then, testing timewarp attacks will be easier. And you can always CPU-mine some stale blocks, and test everything you want. More than that: if you try to mine on top of ASICs blocks, then you have quite high chances of getting a block, if ASIC will not manage to do it in time (which is why I can still mine one or two blocks per day).

The test network was never designed to be ruled by CPUs. If it would be, then you would have different challenges, for example: https://mempool.space/testnet4/tx/4289097ca8ad4f63484b8e278fba5789ee2decbe453e988934c7f9dfe9537ed8

Can you see, what I did here?
You want CPU-mineable coins? Then try to apply OP_SIZE on DER signatures. If you have 56 bytes, then you have to use half of the generator, as the R-value (which means, that your private key will be known), and s-value of 60 bytes (which means, that the first four bytes would have to be zero, so you would need some mining, to get a valid signature here).

And note one more thing: if you have an ASIC, then you can always beat all CPUs, if you have a competition, based on Proof of Work. Because ASICs can just produce much more work, in much shorter time. So: ASIC domination on testnets is inevitable. But: you can still compete with network propagation, so this is the place, where you should focus.

And one more thing: CPU-mined block, even if mined by ASIC, can still be reorged with a CPU. Just be faster, and produce two blocks in a row. It sometimes happens, if you are lucky.

And the remaining blocks go to the high difficulty miners. The high difficulty was a flaw in testnet already, now it looks like there's an ASIC mining 20 minutes ahead in the future too. What's the point of making it impossible for actual developers to quickly mine some testnet coins?

Yes, because https://mempool.space/testnet4/address/tb1q3u8f5899ymkatx69h0n3sw0qpalgwdmrcj80dm is mining 90% blocks. But probably you mined a lot of stale blocks, if you changed the rules, to mine blocks above 2 hours.

Thank you very much. We have been running for 26 hours according to the method in this post. So far, no block has been successful.

I sent you 10.