eli5

What it is the meaning behind L2 "deriving security" of L1? 

In other words:

How does submitting proofs/snarks to L1 makes L2s secure?

Building Solid Goods

linktr.ee/megarocklabs

Imagine a L1 as a constantly growing, huge blob of glass. Every few mins you add a new blob of liquid glass on top, it hardens, and makes it so that no one can mess with the deeper layers without causing a bunch of very visible damage.

Head of Product Serotonin | DΞF

serotonin.co/platform

Now imagine you take a picture of your bank account and lay it on top of the L1 blob. Now when the next blob of glass gets added, your bank account balance is preserved forever. The more time that passes, the more blobs get added on top, and the more secure it becomes.

Submitting snarks etc. to an L1 enshrines them in the 'glass' of the L1's transaction blobs. It doesn't make the L2 more secure inherently, but it does give the L2 a way to prove that a certain balance existed at a certain time by piggybacking on top of the L1's security