It is indeed, thank you.I figure out that the problem was the ":" and just choose another link.
I'll keep that in mind but is so unusual to have problem with links anywhere else that I may foget in the meantime.

These two are improved now.

In this perspective - which is indeed just an idea among others -, I'd suggest keeping the current base price, discount percentage and resulting price, for it's impossible to pass on that fundation, but to simplify the display on your second, sometimes even third set of information.
In your first example, I'll consider two interpretations : A) $10 as a base price in month X, discounted to $2.99 in month Y or B) set to $2.99 as a new base price in month Y. Both are then discounted to $2.97. Regardless, you'd have to display "(A) $10 or (B) $2.99 crossed, A) 70% or B) - roughly - 1%, then $2.97", followed by - roughly - " 71% less than in month X, all time low"... in both cases, since it's the best historical discount. In this perspective, the buyer is only given one information : this is the best possible deal up to now - whether you want to mention $10 again or not is up to you (and available space ^^) - and reminding him about previous discounts or price changes (Y) isn't deemed useful - even though you're still tracking them. I do understand the concept of extra insistence for marketing purposes - "1% better than in month Y !" -, but for those who aren't interested in keeping track of all the historical changes, this isn't absolutely necessary... and saves you precious space.
In your second example, assuming a single interpretation - a base price of €9.99 in month X, a discount down to €2.99 in month Y and a base price change up to €10.99 in month Z -, you'd go for "€10.99 crossed, 71%, then €3.19", followed by - roughly - "2 % worse than in month Y", . In this perspective, you ditch intermediate changes - even though yes, I can sympathize with people who swoon over graphical representations ^^ - to only present the buyer with a simple conclusion : the current offer offered, regardless of tracked price changes isn't the best deal up to now. Considering that it isn't hard to retain only the highest value in a formula - #ExcelPower ! -, a single historical percentage would suit me fine.

Oh and, by the way, check Residual : the two different historical reminders ("50% better" and "all-time low") aren't listed on the same (starting) line...

Can I have a screenshot and exact browser version? They are for me, and should be in all browsers, as alignment happens with align-self: start, which has had global support since 2016.

(That's assuming you mean those two aren't on the same baseline. If you mean those two aren't on the same baseline as the current discount, then yes, that's by design.)

Here you go... On this machine, I have Firefox v115.25.0esr (64bit).

Oh, bother, yes - I understand - I only apply that alignment if there's a highlight in the column, which I check with :has(), which has only been supported since Firefox 121.

I can look into reversing the logic, since I always put the items with highlight on the homepage - but there'll always be some items which I simply can't line out the way I should in Fx 115.

It looks like the lack of :has() support also causes some JavaScript error, as the edition and goodies for Arzette don't have the show/hide toggle which they should.

Well, to be fair, I should have mentioned that I only noticed that because I'm regularly browsing while using my retro gaming rig (with Windows 7 & XP, an older version of Firefox, hardware from 20 years ago at best, etc.). Some hiccups I reported previously didn't show on my work station, which is less of a relic ^^...
It's a good thing to try to take in account slightly older or different logics - it is (or should still be ?) one of Gog's missions after all -, but you probably won't be able to do so in any case. Just do what you can...

Well, considering the previous discussion, plus your project's name, I didn't really imagine that you'd have in mind a full front-end, beyond the scope of an improved search engine. Hence the "extreme" label... Of course, what seems "extreme" to an observer can look pretty "mild" to someone with different ambitions : a common denominator in conflicts on Earth, I might say ^^ !

Mr. Eiii's comment seems a good start, for many Gog users complained about that limitation across the years and I missed myself a few news because of a sudden outburst which propelled the last ones into oblivion a bit too fast... In any case, release informations synergizes directly with your search engine, so it seems interesting.

For the rest, unfortunately, I'm not the best typical Gog user to answer that question, since I don't check the Blog or even the forums often - not to mention the social networks, which I don't use at all -, so I'm afraid I won't be of much help.
I'd say that shortcuts to all (nine of) the "Community" links could be interesting. The social network users will probably enjoy reaching out to Gog's publications there - especially on Twitter and Twitch, for what I've been said even though I don't use those -, the (so-called) Dreamlist is a safe bet - I do click on favorites here from time to time - and I guess (while you're at it) that the older wishlists can't be excluded ; last but not least, the three forums links are obviously used quite often. I can't tell about the blog...

Would trivial information (such as the "About us" sub-links) or the "Help" ones be interesting ? Considering your current approach, I'd suggest avoiding any clutter by merely pointing out at Gog's main page through a single link in addition to your direct shortcuts towards its most important functionalities. A simple "Gog" clickable link before them should be enough in my opinion - as it's not very elegant to force users to open the website through game selections just to reach the website's other parts -, but you need other views than mine...

I don't have issues with that part, but work with lots of people similarly more gifted in personal realizations than in self-promotion, so don't worry, I understand that contrast. I've mentioned Gamesieve in my profile and will continue to do so around me... Unfortunately, since I'm not much of a social network user, you'll need other ambassadors ^^ !

That's actually another good example of common design at least in the fields I work with : you'd never see a full line or space meant to seperate two blocks of information suddenly interrupted, like here, by a "5220 DRM-free games - showing 1–50". We'll all live with the result, mind you ^^, but it does feel weird.

The second issue looks fixed indeed - well done ! -, but not the first one, as it persists in another similar case - which isn't that annoying though. It was simply crash testing situations, by exagerating zoom in and zoom out situations. Take Rogue Trooper Redux : because of his "First time of sale since the last year" (slightly different from Shaq Fu's previous label), on smaller screen resolutions, it will bump all the information under it's main title a couple of lines below. This looked weirder with Shaq Fu, because of its "Included manual" label.

Yes, it's probably the best compromise you'll reach, even if it's not perfect. But, that said, I don't see at all that "Search without applying defaults" checkbox on that version of the Firefox navigator - and that's why I dared to suggest a checkbox actually. Did I miss something or is this another navigator limitation ? One or two checkboxes (lined up vertically one above another, with a label to their right for example) would probably be fine, but I'm not seeing anything for now...

If I understand correctly what you're seeing: At some point the block of ratings doesn't fit next to the "first time on sale in the last year" label anymore, and so jumps below it (but still right-aligned), thereby pushing all subsequent information down as well (leading to a large gap of whitespace underneath the title). One further zoom step, the rating information collapses into a single line and moves left, which looks more logical.
(Ctrl-Shift-M/Cmd-Opt-M enables responsive design mode, btw, which gives you more precise control over seeing how things behave at different resolutions than zooming.)

My issue here is basically the cut-off point for when I turn that rating block into a line. There'll always be pathetic cases of really long titles causing the ratings to not fit, but I don't want to put them onto an extra line too early. Still, I see that at a screen resolution of 1024px wide, 7 out of the 50 games on the current homepage have the jump, so I was probably too conservative with the cutoff (which at default zoom is at 992px), and I'll bump that up a bit. (1168 gets me down to 0/50, so that seems like the thing to aim for).

If you switch to a different country, or exclude some tags/developers/whatever for games which you never want to see, there should be a button "remember and apply these settings by default" underneath the applied filters. (Probably not on your Fx 115 instance, due to the aforementioned JavaScript error.) Press that, and the checkbox will appear. (Plus of course, from then on, when you visit gamesieve, you'll have those defaults applied, including for your searches, except when using that checkbox.)

I do now have a concept for "titles only / exact search" which feels like a full solution: leveraging an autocomplete. When you're typing in the search box, I'll immediately show the top 10 (?) matching games and series. If you click on any one of them, you'll get a search for the specific title of that game / series, with an exact=true parameter or somesuch applied in the URL, causing you to see only that one game / only the games in that series. (I'll probably have to improve GOG's series, but that's a doable exercise.)
I'll need to figure out how this behaves with grouping when the search matches a grouped product, how to distinguish between series and game when the name of the series is the same as the name of the first game (feels like it requires another parameter), and if I want to apply typo correction for the autocomplete results as well. Performance should be good with n-grams, but memory overhead might be prohibitive. Anyway, not quite trivial, so it's going on the backlog until I'm done with lists, which'll give me enough time to properly mull it over, but it feels like a hopeful direction which should fit the requests I've seen.

Um, if such a parameter will exist, shouldn't you just allow users to select it from the options? I'm guessing that any slightly advanced one will just add it manually anyway, assuming it'll work.
But ugh, autocomplete, instant searches, hate that with passion... and always makes me want to type fast enough to prevent results from showing up until I type as much as I intend to, which leads to typos and deleting and annoyance (and also added server load).

Yes, that's it indeed. And thanks for the "responsive design" mode ! Which leads me to new questions about visibility : why duplicating pieces of information such as "Rarely on sale" or "Early access" after a game's title, even though they're already included through available filters, and then adding on top of them other labels ("Recent release" or "First time on sale in...") which could similarly be accessed through filters ? You could save space here...

The same remark could apply to the center number indicator (X DRM-free games with...), repeating the first filter used despite that one already visible enough just one line below - in addition to the filter selector itself. This repetition is even pushed further by removing your presentation ("Gamesieve is an unofficial...") to mention, again (a third time), filters used by commenting those (regarding your price tracking strategy when selecting a price range or displaying "Only showing games which" for grouped products filters for example).
If you were concerned about saving the space it takes or improving the search engine's readability, it is by the way also a bit of information which could (like the Privacy disclaimer) be pushed at the bottom or under the columns - like lots of websites do with their "About..." section for example. And if it was about removing any possible misconception about those filters (but only about the first one selected...), it feels obviously not as complete as other possible solutions, such as a short information panel available when hoovering one's mouse over their category title or over them directly, like you considered in a previous answer.

Moving next, your idea of including a(n arrow) toggle to avoid cluttering search results with various editions (expansions, goodies, etc.) is great, but its behaviour seems a little erratic with expansions. Check Train Life - A Railway Simulator or The Smurfs 2 for example.

Laslty for now, about possible line alignement issues related to font differences, the tag list in the middle looks weirdly positioned by moments, as it seems sometimes aligned on the crossed price and sometimes on the outlined discount percentage. Check the difference between Tomb Raider IV-VI and Knights in Tight Spaces for example.

Indeed and this works as intended, even though I find a bit weird the "reset" options at the very bottom of your lengthy Privacy presentation - especially if that one's bound (soon or later) to be moved at the website's bottom. Not their presence there, mind you - which seems consistent -, but the obligation to reach out for them on a (possibly) regular basis in a different place than the "Applied filters" and "change remembered defaults..." indications. Nothing really annoying though, since it's easy enough to change by removing filters manually...

Yes, that autocompleted concept looks like a possible (but challenging and heavier on ressource) direction to follow, but as far as I'm concerned, a simple "Exact search (words, sentence, etc.)" would be far enough. So feel free to experiment with easier leads too ^^ !
Suggestion : instead of "Search without applying defaults", how about "Search within selection" ? This way, you'd both include selected filters (by default or not)... and allow one to refine his current search in one move.

Oh, and sorry for multiplying feedback. I've experimented with your baby quite a bit now, so comments are starting to pile up ^^... But it's just food for thought of course.

These two things are fixed now.

I understand the feeling, but a significant percentage of my users already seem to expect the basic concept, judging by the number of half finished search terms I see. Typos are also quite common. So if I want to cater to both them and you (not wanting any interpretation/correction of your search terms), that needs some guardrails. I've also been pondering the desirability of a truly "exact" search for e.g. "Fallout" returning just the base game, or also 2-4, tactics and new vegas, and the autocomplete UI allows making that distinction in a way which regular visitors can use. If I can't make it really performant and non-obnoxious, I'll just abandon the concept again. Either way, it's going to take many months yet, and my thinking will undoubtedly continue to evolve.

Because I have many visitors who never use filters at all (same as I have visitors who never search at all, and do everything through filters), and I find it useful myself to highlight such aspects when simply browsing.

These are some of my few nods to (eww) 'SEO' - search engine bots are allowed to access one level of filters, and that level is than added to the main page title. The message about what gamesieve is can be permanently removed with the "X" in the upper right corner.

I only collapse grouped products if there is significant space to be won; the two cases you name both have only two grouped products (and no includes), so I just keep them expanded. I'm strongly considering increasing the cutoff for (probably solely) the first search result when grouping by relevance, as many people search for ultimate editions and the like, and then don't seem to see that they can expand such a section.

It's top-aligned, so yeah, it varies based on the height on each of the blocks. Mostly the baseline will line out with the crossed out base price, but there are lots of cases where it won't.

I would also expect a search to work like this.