Perhaps my language did not clearly convey reality. Let me provide an example. Version X is released on Jan 1. Probably will not ever be installed (see below). Version X.1 is released on Mar 1, might get installed, but not before Nov; for this example let's assume that patches keep coming. X.1.2 comes out May 1 - still not installing it, yet. Jul 1 comes around, if the features in X are absolutely required, and we've had it in the lab, and the customer acknowledges that we are not responsible should the product not perform as documented, we'll install it. Again, let's assume it's a no-go. Sep 1: X.1.3 comes out, we'll aim to install it in November, along with any patches and hotfixes listed in the latest release notes. Bottom line, major release X comes out in January, we'll probably not be running it before November.
Prime example, Windows 2012 began rollout on production servers in January 2015 and we'll have >2000 installs by the end of June. My job is to make sure we never have more than 24 hours data loss on any system, can't do that unless I'm sure the product performs as advertised. In my line of work, there's nothing worse than being told every job/transaction completes successfully but it turns out the data is unusable or empty.
For most of the software I use, a major release X is two digits, such as 5.2, 5.3, and 5.4 or 7, 7.1, 7.5 and 7.6 - each is a major release in its own right, so I'm dealing with 18.104.22.168 or 5.4.5. We're finding that, where in the past the product usually went GA as 7.1.0 with a few hotfixes for corner cases, 7.6 didn't go GA until 22.214.171.124, and the vendor is recommending to skip straight to 7.6.1. When it came time to install 7.5, you needed to install the 126.96.36.199 patch before starting the application. Another example, 5.5.0 isn't GA yet and they're up to 188.8.131.52. One of my colleagues in a different business unit installed RA 184.108.40.206 at the vendor's direction for case resolution, and he was hit by a bug that caused a complete filesystem panic and 18 hour service outage.
Now, to put some real world application to your original statement... Can you name any major release of commercial software that needed 0 or 1 patches? Anything from Microsoft, Apple, any Linux or BSD distribution, Solaris, AIX, HP-UX? Anything from NetApp, EMC, Cisco, Brocade, SAP, CA, Google, Symantec, Mozilla, VMWare, CommVault, Oracle, IBM, McAfee, MySQL, Sybase, Apache? These are the major vendors I deal with. The reality is that software products have become so complex that the vendor cannot test every use case before releasing the product, and invariably GA releases are getting buggier and buggier.
My leadership team fully supports my position - and I wasn't even the one to come up with it. I just live(work) by it. Maybe the leadership team of one of the world's top 100 sustainable corporations are all "fucking incompetent", but I think you're the outlier in this case.
My hard and fast rule is at work, not at home. Yes I used beta software on a personal device that is not used for any primary function. It's not my only tablet and its far from being my only computing device. There is nothing unique on it, I have good, complete, tested backups both locally and off-device, and I thoroughly studied and understood the roll-back process before starting the install. I had relegated that device to lab/disposable status when I did the install. That said, the beta label from PA isn't the same as beta for other products, i.e. Windows. Once PA makes a stable release, that's it, they're not updating that branch ever again and have moved on to other Android releases. Windows OTOH, once the product goes GA, they keep patching it.