As iamnotback has already pointed out in his link Dash does not solve any of Bitcoin's problems related to blocksize scaling. In fact one can argue that Dash actually makes these problems worse since not only the miners, but the masternodes and budget proposals also have to be funded from the diminishing block reward. The only thing Dash has provided here is a temporary kick the can down the road since the Dash network unlike the Bitcoin network does not have full blocks and Dash has an effective 4MB, over 10 min, blocksize as opposed to the 1 MB blocksize limit in Bitcoin.
A simple way to understand the Dash masternode network is to compare it to the United States Banking system:
The number of Banks is around 6700 add to this another 6100 Credit Unions and we are at over 12,000 deposit taking institutions in the United States, vs ~4000 Dash masternodes. Even worse and a more apt comparison needs to be bank and credit union branches since multiple Dash masternodes can be owned by one person. The capital needed to start a bank in the United states ~40,000,000 USD is way less as a proportion of the USD M0 money supply, ~3.500,000,000,000 USD than 1000 Dash for a Dash masternode is of the Dash M0 money supply ~7,100,000 Dash. Do the math please.
The above becomes relevant when one looks at the following:
Instant X. In instant X, 15 masternodes are selected at random to approve or disapprove transactions, by a simple majority. The comparison in the US banking systems would to have 15 bank branches selected at random to approve or disapprove a transaction instead of having a single bank approve or disapprove a transaction. Does that make the ledger decentralized? I do not think so.
The "decentralized" masternode voting process. Here we have the Banks voting in proportion to market capitalization on how to spend 10% of the USD "printed" by the US Federal Reserve. These funds are to be spent on hair brained marketing schemes for example.
In fact, the error in DASH's scheme is that ultimately, the transaction is still written on the block chain. Instant X is nothing else but a kind of certified mem pool confirmation. But if DASH hits the same bottleneck as bitcoin does, and there is a huge mem pool lag with miners picking only the most lucrative transactions out of the pool to make full blocks, your instant-X confirmed transaction may very well never end up on the block chain, just like with bitcoin.
So what is the value of an instant-x confirmed transaction that ends up not being on the block chain ?
In fact, one should push the logic to the extreme, and not write transactions on the chain any more, but just balances. If you trust that 15 random master nodes validate a transaction, one could just as well have them validate the new account states. And then, there's not really a need for a block chain with transactions any more ! Only a shared ledger between masternodes. You might still maintain a small block chain BETWEEN master nodes, where regular "states of accounts" get their hash registered (and only their hash). But the master nodes maintain the balances (which are distributed on demand to the normal nodes, as well as the masternode block chain and the active mem pool, to verify that the latest balance has its hash on that chain, and that the current balance is the right "latest hash stamped balance" on which the mem pool transactions have been applied. Say that every 12 hours, a "current balance" hash is produced by the master nodes, and written to the block chain. So there are 2 entries A DAY on the current block chain. That makes for a tiny chain, doesn't it ! There's the day's mempool worth of transactions which is shared by all nodes, and updated by the instant-X scheme of master nodes, and there's the masternodes' shared "current state of balances" (the delta wrt to the last stamped state of balances), which is the net result of all the mempool transactions.
And that's it.
Very light-weight. Very small block chain. The only thing that normal nodes can do, is download, each time when they are active, the last stamped balance sheet, verify that this corresponds to the hash on the chain, and the active mem pool, verifying that this mempool corresponds to the the difference between the last stamped balance sheet, and the current active balance state (which will become stamped soon).
So no, you cannot verify old individual transactions "when you were not there". Yes, as long as you are an active normal node, you can verify that the balances are updated correctly.
You've just re-invented normal banking.
All transactions HAVE BEEN public. Normal nodes COULD HAVE recorded them (and kept the whole "block chain of transactions"). But the network can forget this for normal operation. A newcomer cannot verify how the balances got to their actual state, but can verify *from that moment onward* that all transactions correspond to all balance movements, and as long as he stays connected. If he "hooks off", he has to trust that the system continues to work that way, and that others to the verifications.
Some normal nodes may keep the full records. These full records are equivalent to the full block chain of transactions. All signed transactions from the recorded mem pools come down to the block contents of bitcoin like block chains. But they don't have to be shared by all nodes. Nodes are only interested in the DELTA of balances when they are active. The don't care about how the balances got there (and if they are interested, they can try to obtain "full block chain information"). But there's no need to do this for current transactions to go on. You can have a daily 10 GB worth of transaction pool, which you can forget about once the balances are updated and stamped.