Wouldn't it be possible to create a MASTER block and start the blockchain anew?

[Tacticat]

If blockchain size is going to be a problem, wouldn't it be possible to create a master block with all currently unspet inputs and start the blockchain anew?

The new block would still be chained to all the other blocks but the client wouldn't require to download them anymore. Why isn't this an option, even if it requires a hardfork?

[anti-scam]

It is an option, but who do you trust to distribute the new genesis block? Look at how much people are whining over a basic change to the client's behavior. The Satoshi's bible-types would explode if anybody tried this.

[jackjack]

It is an option, but who do you trust to distribute the new genesis block?

The old blockchain itself?

• Calculate the resulting hash
• Make sure everybody is ok
• Hardcode the hash

[Tacticat]

It is an option, but who do you trust to distribute the new genesis block?

The old blockchain itself?

• Calculate the resulting hash
• Make sure everybody is ok
• Hardcode the hash

Yes. There would, technically, still be one single blockchain but full clients wouldn't need to download the previous blocks.

[jackjack]

One drawback is that the master block would be huge...
Splitting it in pieces is necessary

[Tacticat]

Multiple blocks then.

If the implementation is changed then blocks wouldn't just hold current transactions but a space could be reserved in them for old unspent inputs, meaning that each new block would hold a certain % of the old blockchain.

The result would be that, block by block, older blocks get deprecated and wouldn't need to be downloaded by the main client. (Obviously they would still exist, but the mainstream client that most of the population used wouldn't need to download the whole chain without giving up security or having to trust a third party).

[anti-scam]

It is an option, but who do you trust to distribute the new genesis block?

The old blockchain itself?

• Calculate the resulting hash
• Make sure everybody is ok
• Hardcode the hash

Then I have to have the old blockchain to verify the hash, completely eliminating the trust-free design of Bitcoin. How is this any different than checkpointing?

[wumpus]

If you create a new genesis block, you're creating an altcoin.

[kjj]

One drawback is that the master block would be huge...
Splitting it in pieces is necessary

Hmm.  Maybe we could split the block into a bunch of self-validating pieces.  Maybe make each one 1 MB or less.  And then we could make a simple P2P network to distribute copies of the pieces as needed.

[nevafuse]

It is an option, but who do you trust to distribute the new genesis block?

The old blockchain itself?

• Calculate the resulting hash
• Make sure everybody is ok
• Hardcode the hash

Isn't this already done in bitcoin-qt?  I'm pretty sure there are at least a couple block checkpoints hardcoded.  Not sure why it still requires you to download blocks prior to that checkpoint.  But to be honest, blockchain activity lately probably doubles every 90 days, so at most even if you did this every 90 days, you'd probably only save half the space.

[Littleshop]

If you create a new genesis block, you're creating an altcoin.

Exactly.  This new master block would not have the appropriate difficulty greater then the real chain anyhow. 

[candoo]

This idea is posted here every 3-6 months.

I was one of the dump people that talked about master block -> view here https://bitcointalk.org/index.php?topic=158467.0

[jubalix]

If blockchain size is going to be a problem, wouldn't it be possible to create a master block with all currently unspet inputs and start the blockchain anew?

The new block would still be chained to all the other blocks but the client wouldn't require to download them anymore. Why isn't this an option, even if it requires a hardfork?

I could never quite figue out why some sort of compiled snapshot freeze could not constitute the foundation for the next period of time

[Sukrim]

If you consider "every block since Satoshi's genesis block" as the valid block chain, you'd need the previous blocks to verify the "master block", which again means that you'd either have to trust 3rd parties for the hash value of the master block or spend time and bandwidth to download the necessary blocks to construct the master block yourself.

As Bitcoin is all about eliminating 3rd party trust, the former is not really an option (there would be simpler solutions than this, if you trust some 3rd parties even to some small degrees), so all you'd really do is to force newcomers into some 3rd party trust scenario or just maybe make it easier to download the blocks that the master block consists of/builds upon as these would be a static set - they would still need to be verified though.

[Bitcoinm]

Why can't bitcoin be coded in a way where you can verify how many coins different addresses have, instead of having to keep the entire history of every transaction.  It's like how miners who find blocks get 25 Bitcoins on their address without those Bitcoins ever having been anywhere before.

[Malawi]

I've been thinking that it should be possible to purge the blockchain, and take away old transactions. There is no need to know that this address sent 14 btc to that address 3 years ago. The important thing is how much there is in each wallet and recent transactions.

[jubalix]

Why can't bitcoin be coded in a way where you can verify how many coins different addresses have, instead of having to keep the entire history of every transaction.  It's like how miners who find blocks get 25 Bitcoins on their address without those Bitcoins ever having been anywhere before.

I suppose this must be smaller than the BC but by how much....alot???, would people start makeing spam addresses?

[Sukrim]

Why can't bitcoin be coded in a way where you can verify how many coins different addresses have, instead of having to keep the entire history of every transaction.  It's like how miners who find blocks get 25 Bitcoins on their address without those Bitcoins ever having been anywhere before.

Do it and create a patch, it's open source (or pay someone to do it).

Sorry to burst your bubble, but internally bitcoin doesn't really know a lot about addresses at all... it cares about unspent outputs of transactions. To refator this would be a HUGE undertaking and as jubalix said, you could anyways just create a lot of addresses with tiny balances instead of a lot of transactions with tiny outputs to spam the chain.