First off, thanks for the opportunity to go back through Dieter's book!
Fatigue does not leave any changes to the structure. Work hardening does.
Fatigue takes place within the elastic limit of the material. Work hardening requires us to exceed the elastic limit. Permanent deformation is required for work hardening to take place.
Fatigue failure of mild steel requires 100,000 to 1,000,000 cycles at 40 or 50 ksi. Alloy and spring steels require tens of millions of cycles for fatigue failure.
Now let's talk about cracks. Microscopic cracks can form in the heat affected zone (HAZ) when welding. (Note the HAZ is not just the bead, but everything near it.) The more carbon in the steel, the more probable a crack will be present in the HAZ upon cooling. When we smack on something with a crack, the crack propagates (gets bigger) by some amount. The amount might be even more microscopic than the crack, but the crack does get bigger. And the bigger it gets, the faster it propagates.
Two ways to help prevent cracks in the HAZ: 1) preheat. 2) post-heat. Both are needed.
Less work to just use those nifty rivets the folk are talking about.