Bitcoin Forum

There is a limit for the number of transactions that can go into a block.  Suppose 206 bytes (if it is like ProtoShares), with a 1MB limit that is about 4900 transactions every ten minutes.  Every transaction must be stored in each full peer on the network.  Moores law might run out as the transaction use continues to go up.  This could make putting all desired transactions on the block chain impossible.   This would force up transaction fees.  Also bandwidth will be more demanding as transactions increase and more people buy into it.  These problems might push out lower valued transactions where sending $100 worth of todays dollars might be too low to use the bitcoin network.

Most people (there are some vocal dissidents) believe the tx limit will need to be raised.  The real debate isn't on should it be but how, when, and to what limit.  The 1MB limit wasn't part of the first version of Bitcoin it was added later as a safety mechanism.  It was and is a denial of service prevention mechanism.   With unlimited block sizes someone in the early years (when solo mining your own block with a CPU and later GPU was possible) could create hundreds of say 1 GB blocks and the blockchain today would be measured in Terrabytes.  TBs of real tx data is a cost Bitcoin will have to deal with.  TBs of "attack spam" would have hindered adoption.  Today with much higher difficulty, miners having to watch the bottom line, and a much larger network in general pulling that type of attack would be much harder and more expensive so eventually I think the (1024KB) training wheels will come off.

Still even with a higher block size it is unlikely that every single tx will be on blockchain.  How much is on vs off chain remains to be seen but off blockchain txs are faster and cheaper (at the expense of centralization).  However while a larger block size means more room for on block chain tx it has its own type of centralization in that the resources for full nodes will be higher.  Higher resources for full node mean less full nodes.  The "perfect" block size would be one which balances these two risks.  If you want to take two extremes: 1GB blocks (in 2013) are probably bad, 1MB blocks (forever) are probably also bad, a better block size is probably in the middle. 

I don't have the answer but I know some questions to start thinking about:
How do we get "there"?
What size blocks make sense? 
How do you balance the two different risks of centralization?
Should it be a new higher fixed limit, no limit, or some dynamic limit?  
How do you manage that in a decentralized network?
Should we look to make a perfect change on the first try or maybe a temporary increase to "buy some time"?

https://en.bitcoin.it/wiki/Scalability

Many problems with bitcoin would disappear if the blockchain were more fully distributed.  
It's just a matter of writing code.

The blockchain if fully distributed.   Every full node has a complete copy.  It isn't possible to be more distributed than that.

We don't need the increase of the limit as it will require hard fork which is dangerous.
The solution will be moving the most of microtransactions out of the blockchain. It will be internal transaction inside trusted third party systems like coinbase, mtgox, sr, etc.
To stimulate that, miners may tighten requirements for microtransactions and force higher fees over them.

Coding it goes way over my head, but we should have an altcoin whos blockchain contains all the microtransactions, and gets reset every x number of bitcoin blocks when the end result of those gets pushed into the blockchain.

With central control?  Sure.  Not sure I see a way to do it full decentralized and trustless though.

Which to all intents and purposes means they won't be Bitcoin transactions, it will be more like Paypal for smaller transactions.

Basically what Death and Taxes said, with the addendum that those loud dissenting voices, when examined more closely, are usually heavily invested in some kind of altcoin that doesn't have a block size limit. Be mindful of that.