>> The current state of software is such that everything depends on everything else.
This is not true. There is a hierarchy of dependencies, yes. But the use of "everything" here is not helpful. There are clearly projects that are used a lot (OpenSSL) and there are plenty of projects that are used by nothing. And everything in between.
>> since the software is free, you can not sell or license it to make money.
again, very not true. RedHat being a prime example, but there are any number of companies selling and licensing open source software. That is explicitly allowed under the various licenses. Mixing the concept of "freedom" and "cannot be sold" is a common FOSS myth.
>> So why do people abandon their widely-used FOSS projects?
For exactly the same reasons they abandon commercial projects. The vast majority of all projects are abandoned, regardless of license. Because _most_ projects are rubbish. If we narrow the set of programs to "widely used", again there is no difference between FOSS and commercial - DOS was commercial, widely used, and abandoned. (For some definition of "abandoned".)
As long as the list posted in the article, there are also many other reasons. Developers lost interest / die / get reassigned / get married / have kids / learn to play golf.
My point - talking about FOSS as a single category is meaningless because it spans too large a ground. It's not like there's a single common cause to project failure, or indeed a single common reason they succeed. Do we even want projects to succeed forever?
>> Therefore, by donating money to developers, you prevent the project from being abandoned.
eh, maybe. For some small number of projects. For a while. Until one of the other items on the list raises it's head. But money is the easiest problem to solve, and in many ways the least consequential.