Deleted

sorry, missed comment, posted 2 comments below

sorry, I totally forgot about it and comments got lost between all thank you spam I got from recent GAs it seems, so anyway just quick reply:

First most obvious one - at least 3 of these avatars can still be easilly found by simplest reverse image search - because original avatars got stored on other websites as well still using older avatar versions. Probably even more can be found if narrowing searches through additional criteria - adding keywords that will narrow searches only to game-related platforms for example.

To go on further, original Lugum's accusation was that only way to solve this puzzle is to know and remember all 1 million users of SG which while may look like this at first sight is in fact absolutelly false. Ofc knowing al people in question would make puzzle extremelly easy and would be a huge advantage in solving a puzzle, but it's the same for any other puzzle - if I make puzzle about theorethical physics someone studying theorethical physics will easilly solve it while other user will have to do massive amount of google searching to solve it. Same is here. If you approach a puzzle in right way you can narrow the amount of work required to solve a problem to a managable amount.

First of all when encounterting any research problem with too big data spectrum, first thing one should do is to try to narrow data the best one can do. So in this particular case let's think about what would be qa profile of user that may question be about. In order to make puzzle solveable Puzzle Maker would rather put more or less recognizeable people into it. No noname bots who never got active on SG, much more likely known users, top contributors, active users etc. What does it mean for us? While theoirethically there are almost 1 million users to check, if we take just this one assumption we can remove vast majority of examples to check. Actually if we narrow it JUST to level, let's say it's our first step, and let's say we're only instrested in top levels so level 7+, it already means that instead of almost one million possibilities to check, we already have just 1775 profiles to check. It means that via just one action we already got rid of 99,8% data which is not important to us.

Let's discuss other ideas of narrowing search. Who is Puzzle Creator most likelly to put in such a puzzle? Ofc ppl she know, ppl she recognize, much more likjelly tp do so than to choose some randoms out of userlist. So now let's think about ways to find out which SG users may steffi know. First most obvious thing to check is steffi's friendlist. And if you'd do that it already be almost bull's eye. Out of 15 questions in this puzzle, 12 users are on steffi's Friendlist. It means that we can already get 80% of answers by checking ONLY 200+ data records. It's 0,02% of 1 million Lugum originally mentioned and this alone can be achieved by hand without any automation, bruteforcing, scripting etc. Other ideas? checking small private groups (steffi is just in 5 of those, all have below 500 users, all missing people can be found there), checking previous SG discussions made by steffi, especially her heated discudsdions about reflinks for example, look for patterns, people she clearly like, people she enjoy walking with, it's unlikely for her to put some hater in her puzzle.

And so on and on, these are just few most obvious ideas, you can come up with even better ways to narrow your data results ofc.

Now let's talk about tools and skills that we need to solve this puzzle or to make it easier. Basically there's only one absolute requirement to solve it (unless you know and remember all people right away) - web archive. But you can find it easilly by googling ways to view past versions of websites etc. We will be using it to look for old avatars of users we want to check. For example to check my old avatars we will go here: https://web.archive.org/web/*/https://www.steamgifts.com/user/zelghadis . From now on it's up to you how you decide to approach the puzzle. All other things only make your life easier and solving faster, but are not required to actually solve the puzzle. You may want an easy way to navigate from Steam Profile to SG profile (for examplre for checkings steffi's friend list) - if you go manual way Enhanced Steam expansion offers one click links on profiles going to SG profile. For automated solution you may use http://www.steamgifts.com/go/user/ID and just grab the data. You can manually search web archive or write a script or use some premade software that will gather data you need (specifically you wanna grab old avatars, which is quite easy as avatar ALWAYS is presented in the same spot of the webpage, is under same webpage element etc. you may compare avatars yourself or use image comparison software to compare all images you gathered. Genereally at this point it's the case of you being tech savvy enough to basically make this puzzle solve itself, but even if you are not - like I said, it's in worst case few hundreds - 1.7k data entries to check, which in given timeframe cqan easilly be done even by hand without using any additional tools. Also good idea would be to go to oldest web archive entry in the first place - if person X has same avatar in oldest entry as he has currently there is big chance he never changed it and you can go to checking next person right away.

So TL;DR:

1. Find cases solveable by reverse image search.
2. Narrow down test data from one million down to managable number. Use common sense ideas - like high level, active users, people Puzzle Creator knows.
3. Use web archive and tools of your choice and available to you depending on your tech skills to solve the puzzle.

I'm also having troubles with Q3 - to me, it could be anyone! =o

14/15