This is a trick with Unicode RTL. The search query in the URL contains the character U+202E RIGHT-TO-LEFT OVERRIDE, followed by the searched-for text in normal (nonreversed) order.
Apparently Google ignores this character when searching for matches (so the matches are displayed normally), but the character appears when the query is reproduced in the search box at top of the screen, so the search string appears backwards.
This character is interesting, one more thing to watch out for in user-provided input. You do your homework and escape HTML in users' comments but if you don't escape the RTL overrides, their effects can "bleed" to other parts of the page, rendering it unreadable or at least hard to read.
Not only that, Google also does keyboard-layout mapping if you forgot to switch your keyboard layout before searching.
For example, if I want to search in Hebrew for "חיפוש בעברית" but accidentally typed in English and typed "jhpua gchcr," (same keys as the Hebrew chars) - I'll still get the results I was looking for [1].
Interesting that you can do this if you put all those url parameters in. But the default google search won't get you these backwards results. if you go to google.com and enter "swen rekcah", you won't get these results.
Also, the url contains &q=%E2%80%AEhacker%20news& which is not hacker news backwards.
Apparently Google ignores this character when searching for matches (so the matches are displayed normally), but the character appears when the query is reproduced in the search box at top of the screen, so the search string appears backwards.