I do not get the following part: "Since each transaction requires the sender to verify two other transactions on the Tangle, more transactions can be confirmed as the number of users sending them increases".
Does this means that a sender is always dependent on future senders to get his payment confirmed (checked and verified by those future senders)?
If so, in theory the system could encounter a problem if 10,000 senders make a payment and thereafter, only 1 person makes a payment each day. It would take a long time before that sender has verified all 10,000 payments before that.
And if no one makes a payment for a long time, all the latest payments (still unverified) would remain unconfirmed.
People making payments do not expect their payment to remain unconfirmed just because no subsequent payments occur. Thats not how the regular banking payment system works either (and for good reason).
Am I missing something?
You are right, this is a problem. But the probabilities that that happens is very low, because this coin is around the world, there are many transactions per second. Think that in this moment many people don't know about bitcoin and despite this the counter in blockchain.info grows and grows every second, then imagine a future where this technology is more used....
One of the current problems with many cryptocurrencies (bitcoin the first) is not the reduction of transactions but the opposite! the scalability!
The problem would become visible if transactions in period 2 would be smaller than 1:2 versus period 1.
And in a scenario that the number of payments remains stable, each payment would be verified by 2 senders, right? Is that sufficient to stop hackers?
Some people suggest that at times when transactions are reduced there could be attacks. Look at this:
However the creators say in their paper that that would not happen. Who will be right? mmm... We need to investigate further.
This comparison between IOTA and Byteball may be helpful as well.