Why keep bringing up this straw man argument? 
Who is using straw arguments here exactly?
There are two straight out lies which are being raised here once again:
Claiming that if you don't blindly capture around the screen then you will have to capture again. Despite that you can capture the entire interesting data without having to capture around the screen.
Saying there is no downside to capturing around the screen. When it does have significant impacts to the trigger/capture system that might cause the user to miss triggers/information.
Repeatedly, over many years. Despite them being completely incorrect. How could its not be clear what I was calling a straw man?
@Someone's flawed argument - user always knows best capture settings for each possible signal in advance and can do perfect capture on 1st try. It is not possible for user to select wrong settings, have doubts, second thoughts or additional insights in the process because all is known in advance. Thus no recapture is needed, ever. Thus "zoom out" has no value and can be ignored. Q.E.D.
If a user has set their scope to capture around the screen then they have explicitly made that choice (and the downsides it brings), which could equally capture the exact same data by using a longer horizontal sweep. Either way the user had to choose what horizontal capture period they wanted.
The comparison is what those two capture methods trade off, how they compare.
Not some imaginary boogieman/strawman argument of but if you set the scope to capture a short period then you might miss something outside that period. Which applies to either method, the same strawman is equally valid to capturing around the window: what if you didn't set the memory long enough? you'd have to capture again!
2N3055's flawed argument - Keysight's "zoom out" should do miracles to get data around already captured on screen data in RUN mode (which in this context is not possible with normal scope). Additional capture around screen occurred later in STOP mode, thus no miracle occurred, thus "zoom out" is not useful.
As explained to you multiple times, the benefit of capturing a longer period with every trigger is different to the "bonus"/extra capture when pressing stop on some of the Keysight models. They are not identical or directly comparable/ The use case which motivated capturing around the screen was the end user seeing something of interest and wanting to see the rest of that specific capture around the screen, not some new capture (which is what is collected when pressing stop).
Keysight's argument - user should get fast update rate in RUN mode because that's useful (extra memory beyond screen slows down things). User should get maximum memory in STOP mode, because that's logical and user would want that anyway (why limit the user?). So memory settings are reconfigured from RUN to STOP mode automatically. This gives fast update rate in RUN mode & max memory in STOP mode (which also allows zoom out if needed) & user is not limited & no recapture is needed (if user would want to look at data beyond screen).
This is where you keep falling down, the additional capture when pressing stop only occurs if there is another trigger after the stop button is pressed. It makes the "recapture" for you, automatic, forced, no choice.
If the user wants to capture that larger period they can do so by setting the horizontal timebase to that at the start.
Other scope brand's argument (like Siglent) - automatic modes are too confusing for users and/or too difficult to implement, so let's fix memory to the same size in RUN/STOP modes. User obviously knows in advance what memory settings to choose. It's user fault if scope is slow (too much memory in RUN mode) or there is no data beyond display (not enough memory in STOP mode). It's user responsibility to jump around like a jackrabbit and change settings for optimal performance in RUN/STOP modes.
The Siglent scope in question (in the thread title) explains how it sets the memory depth very clearly, it is the horizontal view of the screen. There is no jumping around (or other emotive nonsense) or constantly adjusting things.
If the user wants to capture a particular horizontal period they set that before capturing. Regardless if that is by capturing around the screen, or capturing within the screen. You make the plainly false strawman argument that a user capturing around the screen somehow magically does not choose to do so.
So what should be a relatively simple explanation for this new user is now buried in your massive pile of noise.
So I'll repeat the strawman that you are desperately holding onto:
A user needs to select the period they want to capture before they can reliably capture it (the Keysight "bonus" capture on stop is not going to occur in all situations). Capturing around the screen that length/size/period would have to be set in advance, just the same as if it was set by having a longer horizontal sweep time.
Not different, not some unique benefit of capturing around the screen. Exactly the same.