The use of TPM is controversial, they have managed to decrypt disks through the vulnerabilities of the same, and the TPM are like the secure element, in the article obviate this, hence many deny using it.

I prefer my encryption to depend on the strength of my password than to rely on a third party, this has been the Linux approach and should continue to be the case, from my point of view. 

 I found interesting the part about modifying the initrd image. I've done that in the past to save data from a broken system.

Just somehow never occured to me, that you can do malicious things like altering the system after the user decrypts the data. 

 Check out who writes this article, they are Microsoft guys.

Read this article, inside it there is a link to yours.

https://unixdigest.com/articles/the-real-motivation-behind-systemd.html 

 I don't think they are wrong about this attack vector tho...

(Btw. I'm not advocating for or against systemd. Users will decide what's best for them. E.g. for my line of work things like systemd-growfs/makefs are very valuable. It doesn't mean I have to have them on my laptop.)