Hello gogtrial34987!

Thank you for adding this feature!
I already made use of it, and it works really great.

By the way, recently in the Regional Locked Games thread, Cavalary posted an api-call to list all blocked products of a region (see here).
Looking at the output of this api-call, I realized that the respective products are listed by their product ID.

I am not sure, if there is an easy way to get the product ID of a given game/product or associate the corresponding game/product to a given product ID number.

Would it be possible to add two related features to your gamesieve:
- First, allowing to search not only by name/title but also by product ID number.
- Second, add the product ID number(s) as extra information on the search results. (This could be by default hidden, only being displayed by either a general option or on a game by game basis in the search result.)

Maybe these features could be of interest to other users, as well.

Thank you again for your continuous efforts regarding the very helpful tool you created.

Kind regards,
foxgog

Glad to hear it! :)

All GOG APIs use that product ID for everything, so yes, internally I make heavy use of it, and the title of a game is effectively only "for display purposes".

Technically it is trivial to do this, but I'm afraid this ability would cause gamesieve to be used as an API endpoint itself (as I've seen happens with gogdb's search and product pages), which is a responsibility I don't want. Is your purpose purely to discover which games are banned in your country? Because I'd rather add some utility pages for that purpose. (I'm already planning to extend the list of countries for which I gather this information.) Or what is the functional thing you're trying to accomplish?

The additional annoying thing with searching for IDs is that there are 7 games with IDs 4-10, where everything else has 10 digit IDs. And there are also games where numbers in the title are important. Regular visitors would not understand why searching for "6" would give "MDK" as the top result, rather than e.g. Tropico 6 - but if it isn't the top result, searching by ID feels flawed.

The product ID is already in the HTML for each item, as the id attribute on the li element (or for grouped products, on the dd element). You can make it visible with a bit of user styling (using stylus, or in Firefox even without an extension by creating [profile dir]/chrome/userContent.css ):

@-moz-document domain("gamesieve.com") {
ol.results > li[id]::after { content: attr(id); position: absolute; margin-top: -1em; }
dl:not(.incl) > dd[id]::after { content: " " attr(id); }
}

Unfortunately the displayed text can't be easily copy/pasted, so the usefulness is limited. Basically the same question here: what is the actual thing you'd like to know / see / be able to do, for which the ID would be helpful? Because I'd like to see if I can solve that functional usecase, not just for you, but for everyone.

I'm currently working on adding support for filtering and sorting by rating, but also taking steps to design what I'll do for wishlists (e.a.), and for that I would like some feedback.

I want priorities for wishlists. Originally I thought I'd make this completely variable, and up to each individual user to determine how many priorities they'd have, and how they'd name them (as a step toward all kinds of other public lists, aka gog mixes) - but I've started to think that have a stricter framework would be beneficial, as I also want the ability to filter by (public) wishlists of "your friends", alongside with the ability to see their priorities (if they choose to share this).

(Again, all of this is still super long-term. Don't get excited, as I have no idea yet if it'll really work out. I start by thinking up my ideal world, then see what that requires on the technical side, and then I'll pare it down until it's doable.)

As an initial take, I'm thinking:

Wishlist priority:
- high
- medium (default)
- low

Hide list:
- hide (default)

Owned:
- on gog
- elsewhere

What I mostly want to know is: Is 3 levels of priority enough for wishlists? If you believe there should be more, please explain a bit to help me understand the mental distinctions you make between different levels?

Is there value in having distinctions in the hide list? I was originally thinking yes, but after some thought realized that they'd need to be exclusive categories - so any distinction I can think of (like "bad", "offensive", "doesn't meet my system specs", "AI", "multi-player DRM", ...) would be better served by a separate layer of user-tagging / gog mixes, as multiple can apply at the same time. (But open to hearing ideas for exclusive categorization you'd like to see there.)

I'm not going to try to be a game management service, so don't want to split up "Owned: elsewhere" into the various competing services, but otherwise same deal: open to hearing thoughts.

Your initial plan is just fine for my needs.

Another feedback question, now on ratings. I'm pondering two interconnected things:
- In general, I find ratings by verified owners much more helpful/trustworthy, but I don't want to discount regular ratings altogether, particularly for games with few ratings.
- Sorting on rating becomes less useful due to niche games with only a single "5 stars" vote. I can offset this by adding a couple of dummy "3.5 star" (?) ratings, to basically weigh down niche games and put them on a "level" playing field with more popular games.

My options:
1) Separate sorting and filters for both verified ratings and total ratings.
2) Sorting and filters for a weighted average between the two. Games like Dustborn (15 verified ratings with a 4.5 star average, versus 358 total ratings with a 1.8 star average) make a proper ratio hard to determine, though. 20:1 feels the minimum for that (resulting in a 3.9 rating), which seems excessive for pretty much anything else. Fancier would be a 10:1 ratio, except for games that have more than 10x as many unverified ratings, where I'll clamp the total somehow. At some point I have to accept I can't fix reality, though.
3) All of the above. This feels excessive, and is basically a cop-out failure to determine the best course of action, so I'm not going to do that.
4) Sort by weighted average as in 2), offer separate filters for both "pure" verified and total ratings. <- Leaning toward this.
5) Sort for both pure verified and total ratings, filter for weighted average as in 2). (I can't think of a usecase where this'd be desirable; only adding it for completeness sake.)

A) Accept that niche games will top the sorted list.
B) Add "a few" dummy votes; niche games will still rank relatively high, but not unduly so.
C) Add "a lot of" dummy votes; sorting by rating basically becomes sorting by popularity.

I'm leaning to 4) and B) (writing this down helps) :) - but basically want to check for strong opinions and/or usecases I'm not considering.

I am leaning towards 1) and D).

I believe any weighting will result in unexpected outliers that may confuse users. Regarding 1), I would make total the default and verified the alternative.

D) Consider a game unrated if it does not have some minimum number of votes. GOG used to have this (or perhaps still does), with a cutoff of 5 votes. I would suggest a higher cutoff.

Regarding the first questions, that can work, though most of the local copy of my wishlist has "watching" as "priority", as in I just have those games there to keep an eye on them, maybe track price, maybe check how updates are going, so basically I didn't decide that I'm not interested, but I'll try to see whether I'd actually want them if/when the opportunity to readily get them will present itself.

Regarding the second, I think by default the actual ratings should be shown as they are, allowing the user to select between overall and verified only and also displaying number of votes. Only after that, if/when you can get around to it, you may have something more complex set up. As for what that could be, users who want a weighted overall score could be allowed to set their own weights for verified and unverified owner ratings, and to cover for the low number of votes you can use the survey error margin calculation, 100/sqrt(number)%, taking the result that much towards the middle score of 3. For example, if there are just 5 votes and the real average is 4.8, you'd get 4.0 (sqrt(5)=2.236..., 100/2.236=44.72..., 4.8-3=1.8, 44.72% of 1.8 is 0.80, 4.8-0.8=4.0). On the other hand, for the Dustborn example you gave, it would hardly change the total score, adding a mere 0.06 to it, 5.29% of the difference of 1.2 to 3.0.

Oh, absolutely. I definitely won't tweak what's displayed - it's only the sorting order where I'm thinking about a hybrid approach.

That's unfortunately never going to happen. It's too much a power user feature, and for performance reasons I want to (effectively) calculate sorting order in advance, instead of doing this at runtime.

Thanks, I wasn't actively familiar with that one!

Yeah, that's very much a risk with doing anything complex.

Any reasoning? The more I look at the games where there's both a strong difference and a significant number of verified ratings, the more I believe that verified is the only worthwhile one.

This was helpful, in two ways:
1) making me realize clearly that I don't want this. If you're sorting by rating, there's some rating below which games are "bad". Unrated game should appear above those games, not below them. I don't know exactly where the cutoff is (3.0?, 3.5?), and it's probably different for everyone - but above is better than below.
2) The idea of a cutoff is still helpful. I'm currently pondering variations of this concept: If there are (cutoff) 10 or more verified ratings, use only verified ratings to sort. If there are fewer, but more than 10 total ratings which are within a margin (0.5?) of the same rating, use a (weighted) average of them, otherwise (if there are very few votes in total), (also?) use dummy votes / survey error margin calculation.

Because I'll have filters for the pure values as well as sorting, you can still really dig for specific traits (e.g. show me all the games with a verified rating of 4.7 or higher, but 3.0 or lower total (or vice versa), with at least X number of (verified) ratings.

Depends on the game. If you look at classics the difference between the two is minor, if there's any difference at all.
I'd say the problem is mainly for new games that are noteworthy, for better or worse, so few have played them yet but many have opinions based on news, hype, discussions...
And in case of games newly added on GOG but which aren't actually new, initial reviews are likely to be from those who played them elsewhere, and those are useful.

1) I assume that verified owners are those defined as such by GOG --- i.e. by the same company that believes I own all the Witcher 3 DLCs, but not the base game.
2) I see no reason for people who own the game on GOG to be better (or worse) reviewers than those who own it elsewhere (or in some cases, do not own it at all).

Based on the first 2000 product ratings I've gathered, 16% of verified ratings are identical to total ratings, and 56% are within 0.2 points. For all of those, it effectively doesn't matter which one I sort on. It's the 5% of products where the two numbers differ by a full point or more which I'm trying to solve, and for that group, every single game I've looked at has the non-verified ratings being significantly worse than the verified ratings - on the negative side this shows as anti-woke, anti-vn, anti-adult etc ratings (basically rating the game genre, rather than the specific game), on the positive side (much rarer, but there) it almost always smells strongly of astroturfing/ballotbox-stuffing. (E.g. Lone Survivor: The Director's Cut has 9 verified ratings with an average rating of 2.9 of which 7 left a comment, versus 179 total ratings with an average of 4.0, of which 16 left a comment (including those 7 verified ones). (The normal ratio of verified to total is 1:2 through 1:4 depending on age of the game, not 1:20.) Or EDENGATE: The Edge of Life has 10 ratings; 6 by verified owners of which 5 left a comment, resulting in a 2.0 rating, 4 by unverified, none of which left a comment, but simple math indicates they gave three 5-star and one 4-star ratings.)

Obviously I can only judge reasoning where people leave comments. It's very much not the case that a verified review is by definition good, and I take your point wrt GOG's capability of determining what's verified - but on the whole, where it matters, the quality difference is notable to such a degree that I feel justified making this decision.

Fair enough. You make a compelling argument.

I failed to stop some annoying scraping bots, so I'm now suffering from the already feared combinatorial explosion where they're effectively trying every single combination of filters, 4-5 levels deep (so that's 1,000,000,000,000,000+ URLs for this bot to try to visit). It's mostly coming from residential addresses.

I've resorted to some heavy-handed measures, including blocking huge swaths of ancient browser versions. Firefox 115 and a few others which I know are in use by real people are safe, but if you use another ancient browser, there's a non-zero chance that you'll be blocked. If you are affected, please let me know your exact browser version and/or IP address, together with the rough time when you tried to visit but were blocked, so I can craft an exception.

I have been struggling with this on my website for close to a year now, working along the same lines that you seem to be taking. My experience is that these kind of heuristics ultimately do not work. I was able to block more than 95% of bot requests, yet over time the absolute number of "legal" requests just kept growing. It's as if the blocking heuristics acted as evolutionary selective forces, with bots blocked "dying out" and bots allowed through "proliferating".

Earlier this week, after an especially aggressive 24-hour onslaught had crashed my website for the third time in six months, I gave up and password-protected my search engine. From now on, my users can still access static pages as before, but when they want to perform a search (which is the primary usage of my website), they need to "sign in" using a username-and-password pair (which I post in the open on the search engine page).

As soon as I implemented this (three days ago), the number of bot requests dropped from thousands an hour to exactly zero.

You might want to familiarise yourself with the intricacies of configuring your web server (I use Apache) for password protection, in case your day comes too.

Ratings are here now:
- Displayed for all games.
- Sorting options for number of ratings (effectively a popularity score) and weighted rating (which is just the verified rating when there are more than my cutoff of 20 verified ratings, and below that a hybrid weighted score between verified rating, total rating and the true average of all ratings (very close to 4.0), with the verified rating counting heavier the more verified ratings there are. It feels too complex to make much sense when just eyeballing it, so I can imagine still simplifying it if I get too many complaints about it - but it works well for every usecase I've thrown at it).
- Filters for total/verified rating and number of ratings (total/verified). The combination of these makes for a great way to find the most controversial / heavily brigaded games on GOG. (The inverse works for surfacing a few shilled games, but also makes it apparent that in the early years of gog, all reviews were unverified, and they haven't retroactively marked reviews by owners from those days as verified.)

I also fixed a bug where the prices for non-US countries displayed a wrong "previous time" date for games at their all-time low, when the previous time happened before I started gathering non-US prices.