Did you read the article? Here's an extract of the key part:
There is an entirely different debate about how much or even whether the block size limit needs raising, but here I will solely be doing my best to explain one of the ways a future block size increase could be achieved.
To begin with, I need to explain the difference between a hard and soft fork.
It relates to backwards compatibility. Imagine you and many others are playing an online game where you all roam around a virtual world together collecting flowers to score points. You’re all trying to achieve a high score.
One day, the game releases an update, and adds mushrooms to this world. Just like flowers, the mushrooms can also be collected to score points.
How this upgrade is implemented can be compared to a soft fork and a hard fork.
If the upgraded version of the software is not compatible with the old version, you have a hard fork. In this situation, anybody who updates their software will be in a new world, we’ll call it mushroom land, and anybody who has the old software will be in flower land. The people in mushroom land know they arrived there through an update. Back in flower land, people may wonder where all the other players disappeared to, but will continue to continue playing amongst themselves, not realising that everybody else is playing an updated version of the game, and that the high scores they’ve worked so hard to achieve will be lost when they eventually realise that they need to upgrade to mushroom land.
If the upgraded software has backwards compatibility with the old version, you have yourself a soft fork. In this situation, maybe people with the upgrade can see mushrooms, but to those who haven’t, the new mushrooms just look like more flowers. Crucially, people with the old software can still play with those who have upgraded, and even though the world looks a little different – ultimately everybody is still playing the same game.
So is there a third way?
Maybe there is. Let’s imagine that for technical reasons there is no way mushrooms can be introduced, even as flowers, without a software upgrade. Perhaps though, all players can participate in the same virtual world, except that those with the upgrade can also see mushrooms, while those without cannot. Crucially, people are not playing a completely different game, and when someone in flower land does upgrade they can take their score with them, and start collecting mushrooms without losing out.
How can this be achieved in Bitcoin?
Instead of just losing points, a worst case scenario for a Bitcoin hard fork would be people losing money. Whatever happens, there must always be a block produced with 1MB or fewer transactions, otherwise Bitcoin will hard fork as old software will not recognise the new block.
So how do we prevent anybody losing money? We create a mushroom block. This mushroom block can be larger than 1MB. The existing 1MB block remains in place as a legacy block. Transactions that make it into the legacy block can be viewed by everybody on the network, those in the mushroom block can only be seen by those who have upgraded.
Once any Bitcoin has made its way into a mushroom block, it can not be viewed or spent again by software that has not been upgraded. This isn’t quite a hard fork though as people on the old software will still be participating in the same network, they just won’t be able to see all the transactions that are actually taking place. If they try and spend Bitcoin that appears to be theirs on the old software, but actually has been spent on a mushroom block and belongs to someone else, all that will happen is that the miners will reject the transaction and they will not be able to spend it.
Ultimately, this is not ideal, and while it would no doubt create confusion, it would not create a complete hard fork and should not lead to anybody losing Bitcoin.
The only people who would actually need to upgrade for this to take place would be the miners. Technically speaking, they could have already done this and none of us would know. Since mining is far more centralised and dominated by a smaller group with a huge financial stakes, it is likely that such an upgrade could be achieved rapidly, once the software is ready of course. For everyone who hasn’t upgraded, Bitcoin would just become increasingly less useful until they did so.