Member Ranks

[euler]

To help the community feel more alive and to provide an additional level of engagement as you journey through the problem set, we are proud to announce the introduction of ranks. Those of you who were around in the earliest days of Project Euler will have a sense of nostalgia.

Members with public accounts will now be assigned to ranks: Tetrahedron, Cube, Octahedron, Dodecahedron, Icosahedron, and the top 100 members will be placed among the legendary Spheres.

It was felt that a more meaningful measure of progress than just the raw number of problems solved was needed, which is simply a measure of the number of correct answers entered. Consequently, solve counts will no longer be visible in any public tables and ranks will appear in their place. Although progress through the ranks is related to the number of problems solved, it factors in thoughtful engagement with the problems and the length of time you have been an active member of Project Euler.

For newer members, you will be assigned to the rank of Tetrahedron once your account reaches an established status. Thereafter, you can see your progress from one rank to the next on your Progress page. Please note that your account must be switched to Public status to be ranked and ranks are updated on a daily basis.

We hope that you enjoy these changes and as always if you notice any issues, please report them here.

[universalset]

Someone people in the Discord server noticed that the octahedron and icosahedron images are reversed.

[yourmaths]

If I am not mistaken, this update is intended to discourage cheaters/AI users/solution publishers.

Respectfully, though, I think it is a mistake to remove the number of problems solved. Many people find the increase in the problems-solved counter a major motivation in completing problems. Moving to the next level can take a year or more of problem solving, and increasing ranks would take even longer. Much of the feedback/reward from completing (what can in many cases be very difficult) problems will be taken away with this change.

Why not report both rank and problems solved?

[marecik16]

I am not happy with the update.

After my return to PE, the biggest motivation for me was to reclaim the 2nd place in polish ranking (the 1st place is practically impossible...). It was very useful for me to see how many problems were done by the guy I am chasing: well I still need 30 more, it will take me at least a few months and many brilliant "eureka" moments, but the goal was clear. Far far away, by clear. Now I quite lost this perspective.

[abiramia]

Despite my humble ranking, I would like to express my agreement with yourmaths and, especially, marecik16.

Chasing the next spot on my national ranking has been a great motivator and feels much more tangible than improving my polygon ranking.
Perhaps location-based rankings could still display and be determined by the number of solved problems?

Either way, I am grateful for the developer's efforts to keep things fresh and motivating for newcomers. Thanks!

[Swistakk]

In the announcement in the News section we can read "To help the community feel more alive and to provide an additional level of engagement as you journey through the problem set (...)" as well as "solve counts will no longer be visible in any public tables". These two sound very much contradictory to me. As for the removal of awards, I can somewhat see the reasoning about trying to cater to more casual users (even if I very much don't agree that it caters to them at all), but for the removal of solve counts I simply cannot comprehend what was the rationale behind it. The reception of removing the solve counts was unanimously negative on the Discord community and everybody is bamboozled on what was the thinking process behind it. It caters to no one - neither new nor experienced users.

Let me break the argument into more specific points:

Removal of solve counts:
This is the worst part. This solve count is the most direct motivator for pretty much everybody and the measure of progress that everybody was looking at and comparing themselves with. Removing it is a very direct hit to all users motivation and engagement. As I understand, their removal is tied to the next point as I don't see anything wrong with displaying it other than it would have looked awkward if people on the leaderboard would not have been sorted by it.

Introducing a new progress measure:
The leaderboard measure has to be a stone cold, very tangible and transparent public formula. Whatever "thoughtful engagement" means, taking it into account is very vague and questionable. I guess "length of time you have been an active member of Project Euler" can probably be rigorously defined, but nobody wants to see that in the leaderboards formula and it also disincentivizes solving multiple problems in a short time span. You may create a separate leaderboards that reflect engagement/activity, but it should not be a part of the main progress measure that should have stayed unchanged (the only change to it that I'd find reasonable would be to take into account difficulty of the solved problems). Akin to Codeforces "Rating" and "Contribution" leaderboards that are very much reasonable on their own, but should never be combined.
Also, this measure is very much not understood by me. I have a friend that solved ~the same number of problems as me (4 more), but their mean difficulty (~15) is visibly lower than mine (~21), he has less thread posts than me, less kudos, but he's still 6% closer to the Sphere than me. The only thing he might have going for him is the number of unique days he solved at least one problem on/visited the site, but that's a very questionable thing to take into account at all.

Coarser ranks:
We had 985 possible values of solve counts and 40 possible values of levels, now we just have 6 ranks. Looking at leaderboards seeing people categorized into 6 buckets without any more granular measure displayed is a greatly unsatisfying sight for me. The fact that I can hover my cursor over ranks icons and see the progress % to the next rank somewhat helps, but only partially. And having only 6 ranks means that the jumps from one rank to another just became astronomical. With levels every 25 problems solved, there was always some goal nearby. Now, users entering Icosahedron with ~450 problems solved face "wow, to advance to the next rank I need to solve 400 problems more". And for the Sphere users it seems that their progress will not be reflected in any way neither publicly nor privately (except for the number of problems solved on their own progress tab), which is quite demotivating and especially bad given that 850 problems are arguably closer to 0 than to 984. And new users have to solve ~150-200 problems to get out of the "just registered" rank.


I got hooked to ProjectEuler two years ago, but recent changes definitely sway me away from the site and I don't think I would have gotten hooked if the site two years ago would have been in the state it is in now. It's obvious that you have good intentions, but it seems that your view of what's the best for the site is very different from the view of the community that you want to serve. My request would be to bring back displayed solve counts and make it again the main leaderboards measure. For me the ranks can stay in some form, but not as the _main_ progress measure.

[Patryqss]

I'll allow myself to quote gor3n from the Discord server:

so PE is not run by community, but by fear of 20 cheaters...that should be most sobering thought...very sad

And I think this is the most accurate sentence that I've read as to what's happening to the PE in the recent months

[abiramia]

I'll allow myself to quote gor3n from the Discord server:

so PE is not run by community, but by fear of 20 cheaters...that should be most sobering thought...very sad

And I think this is the most accurate sentence that I've read as to what's happening to the PE in the recent months

I find this wording a tad bit harsh and rather imprecise.

The maintainers identified the need to implement changes to keep things exciting for newcomers and minimize the impact cheaters have (apparently, correctly so, but I have not delved deep into what is actually going on).
If we as community believe that they have over-corrected, we should voice it, but in a constructive way - especially given that all the work is done by volunteers. Just my 2¢.

[Anonymoose]

After not solving any problems for a few months and recently checking in, Project Euler is completely different! As mentioned by other posters, I understand everything here is provided free of charge on a volunteer basis and I am grateful for all of the effort put into maintaining the Project.

This paragraph may be off topic, but I mention it for contrast against another recent major change - the massive reevaluation of problem difficulty. While this unfortunately means I will never realize the emotional satisfaction of solving a 100% difficulty problem, and seeing the Problems Solved By Levels chart shift around after a poorly timed page refresh can be jarring, it has benefits. Relative difficulty of problems fell apart after the first couple hundred problems, but this new, dynamic difficult system tries to address things like that which had been underlying issues for years. Overall, my opinion is the problem that the new difficulty tries to solve is much more clear and apparent and the shortcomings can be lived with.

Now, on to the ranks and removal of levels. I'll be the first to admit I'm hardly competitive here. I won't be topping any leaderboards any time soon and that was never really the goal for me. If I have enough free time and functioning brain cells at the same time, with some luck I can manage to gain one level in a year (Level 8 after about 13 years, clearly I've been averaging less!). But dammit, gaining that one level felt good. All of that time and thought and effort finally culminated in a clearly defined stepping stone and felt like an actual achievement. When I started and got through the "quick" earlier problems, even Level 8, where I stand now, looked like a monumental task. Even if it's not much in comparison to all the Level 30+ solvers around here, it's still something I couldn't have dreamed of in 2016 when I hit Level 4. Even though it's only just a comparison against myself - which may or may not be the only competition that really matters - I know that past me would be proud of where present me got to. Seeing the progress of Level 5, Level 6, Level 7, Level 8 marked major steps along the way that I could point to as unmistakable markers of improvement.
Before, I was Level 8. Now, I'm ... an Octahedron? ... with 27% progress toward Dodecahedron? ... What? What does that even mean? There's not a lot of intellectual meaning or understanding of progress for me there and even less of an emotional component.

If we want to insist on keeping the ranks, fine. The limited $100n^2$ spots for each rank above Tetrahedron seems like a competitive and cute idea (even though it's curiously not mentioned in any of the News postings about it?). I'd rather revert back to the problems solved count and level system, but I'll give some potential ideas for having the systems coexist:
- Show all of the data points: level, problems solved, and rank
- Bring back the old problem count and level projecteuler.net/profile/\$username.png and have a separate profile image with rank accessible under a different URL
- Add a setting to the Account page for how projecteuler.net/profile/\$username.png is displayed on a per-user basis and let each user choose if it shows problems solved+level or rank (or both?)
- Add a setting to the Account page or Statistics page to choose between displaying level or rank

[eskimal]

I feel a little bit of sorry every time I discover a new change in PE, but I also have a long experience running a website and I know that, no matter how good the reasons for making a change may be, users always will complain.
I have mixed feelings about all the recent changes in this season: I've felt discouraged after discovering there were some cheaters, and for the disappearance of levels. On the contrary, I felt relief being freed from trying the unreachables (for me) Gold Medal and Master of the Archives, and I guess I will start feeling fear of losing my precious Sphere rank in the near future.
Ultimately, I don't think these changes matter much; in any case, I will probably always have some of PE's problems on my mind

[euler]

Please note that the following is not so much a discussion, but rather for information.

The primary goal of Project Euler is to provide high quality problems that serve to inspire, educate, and challenge our members. The threads should be an important part of that experience, allowing members to share their insights and further each others learning.

For this reason, a number of the team were keen to have all public tables removed, as they have been the source of more trouble than they are worth and have distracted us all from what really matters. So what we have provided is a generous compromise. Yet, I genuinely believe that this compromise is in fact better than anything we have had before.

As has already been explained, solve counts, and consequently levels, are part of a personal journey and have no place in the public arena. Levels are still very prominent on a member's own Progress page and can be shared with friends. But with reference to showing this information publicly, as the problem set has grown to the size it is today, the sheer number of problems makes a journey to the highest levels demotivating to newer members. Ultimately, solve counts put too much emphasis on the destination rather than the journey. If members choose to speedrun the problems to get to level 38, then that is their choice, but we do not design problems for that purpose. They are to supposed to be pondered and enjoyed, and in doing so, you show the greatest appreciation possible to their creators. Just take a moment to think about the time and the energy that the team devote to developing just one problem.

Ranks are not levels!

Ranks provide a more balanced measure of how members tackle the problems. I appreciate that not knowing the details of the mysterious algorithm behind them might be frustrating, but anyone who has been around for some time cannot fail to agree that those who are found in the highest ranks have rightly earned those places. They just seem to get it right and there will be very few surprises of who is there.

But more importantly, due to the way in which continued attention to the problems and the number of members tackling them diminishes over time, it is those who are most persistent who will be naturally rewarded with rising through the ranks, without having to solve 800+ problems.

However, for those who have been around for a long time, levels will have become part of the driving force, so when it comes to ranks, a change in mindset is probably going to be required.

The spheres represent the top 100 members of all time, and clearly the rarity of that achievement is reserved for the absolute legends. Even getting to the icosahedrons is an outstanding achievement. You are essentially ranked among the top 500 members of all time at Project Euler. Getting to any rank is something that should be celebrated, and that is the purpose of ranks. Giving a fairer acknowledgement of the members who have persisted and succeeded.

One final point to consider. The ranks appeared suddenly. Your position there did not. Those ranks measure your long term success; most likely years of dedication. Consequently, you should expect your progress/movement to reflect that. And for some, your position may eventually settle. Not everyone is going to be a sphere. That means you may have found your place among the all time greats of Project Euler.

Those of you who have already achieved a ranked status should pause and take a moment just to consider how hard you have worked to earn that position. That is what we are celebrating along with you.

[carlsen]

i get that the spheres are meant to celebrate persistence and overall achievement - and being top 100 all-time at PE is definitely not something to sneeze at.

but i honestly think the current metric is too narrow. it reflects only total solves, and that doesn’t really capture what someone solved or how rare those solves are.

for example, someone could focus heavily on very hard, legendary problems - and that effort isn’t really visible in a pure count metric. solving several problems that relatively few people have done all together reflects a different kind of dedication than just accumulating volume.

also, take Gennady as an example: he has 38 gold medals, an astounding achievement, yet this metric doesn’t seem to highlight that at all. it kind of shows that the system explores only one degree of freedom.

maybe a richer ranking could combine categories such as:

* total number of solves
* number of solves among the 50 hardest problems
* number of top 50 finishes
* number of top 10 finishes
* number of top 3 finishes

and carefully fashioned weighted combination of those would give a much more nuanced picture of excellence - rewarding both depth and breadth.


gor3n

[hacatu]

I don't think this is a good update, but it seems that feedback is unwanted, so I will just repost a couple of my comments from the discord and then speak no more of it.

I'm not deleting my account yet, but I'm thinking about it. More likely I would just set it to private. I still enjoy solving problems, but the last changes have been one bad change after another. Unless the changes get reverted, there's no reason to have a public account anymore. Someone else can become an """icosahedron""" instead.

I already said my piece on the awards update. But to restate it simply, solving all the problems is not a possible goal for almost anyone. [To be clear, I do not think that I can solve all the problems.] Obviously, people who do solve a lot of problems can set their own smaller goals and don't need the motivation of awards, but they still serve as a good intermediate goal. Why remove awards that are "too difficult"; are we also going to remove problems that are too difficult?

The big issues with the ranks update are 1) a lack of transparency and 2) top 100 is DRAMATICALLY easier to achieve than max level, but ranks are RIDICULOUSLY far apart in general, so getting from Dodecahedron 100% to Sphere could take years. And Sphere is the "top 100 of all time", but Min 25, Muthu, and LucyHedgehog are not Spheres?

the sheer number of problems makes a journey to the highest levels demotivating to newer members. Ultimately, solve counts put too much emphasis on the destination rather than the journey.

Is this true? Have new members said this? I care a lot more what new members think thank what the top 100 ish think ... but it's also worth noting that there are only a couple thousand active users, and I don't see how putting fewer steps along the journey is MORE motivating

I guess you are saying that you don't want solve counts to be public information at all, which is kind of understandable, but I think that most active users feel the opposite way

Project Euler is not a democracy, so you can make whatever changes you like, but given the number of active users is so small (just look at the number of solves on any recent level 0-3), I think following what users actually want would be good

I've definitely not been solving problems as much

[euler]

I love Project Euler as much today as the day I created it almost 25 years ago. Nothing I do is out of malice, but out of necessity and a desire for this place to be better for everyone and I work tirelessly to achieve that goal. I read every post and carefully consider every comment made. So my comment was not saying we don't want feedback. It was making it clear that nothing can change the outcome. As a team we have already spent far too long on this issue and what we finally settled on did not happen on a whim. Like many others I am also disappointed that tackling this issue became a necessity in the first place, and if I were an active problem solver I also might feel disappointed with some of the changes. But we earnestly believe that this is the best compromise.

Without revealing too much, for obvious reasons, I can share that the system which monitors solve patterns has been many years in the making and performs a relatively sophisticated analysis. Working with masses of data I have constantly analysing and refining it so that it gives a reasonable benefit of doubt and a fair balanced verdict with a high degree of certainty. Does it get things wrong? Yes, but the outliers for the honest members are so few where they exist and those rare anomalies are in almost equal proportion for all members. So essentially all honest members are impacted equally, albeit negligibly. Do dishonest members get past? Yes, but nowhere near as often. Due to the effectiveness of the system this metric has provided the perfect solution to measuring positive engagement and has allowed us to create the ranking system. However, publishing those raw values would undoubtedly lead to reverse engineering which would both undermine the effectiveness of filtering out unnatural solves and gamifying the situation, and it may even lead to some members using the data to judge others. Plus it would inevitably lead to numerous appeals and objections from those that focus on the tiny details, citing things like "On such and such a date they had to do X, Y, and Z, that led to this and that, so could you please override it on that occasion?" This would add masses of work for the team and lead to more distractions from our primary focus: problem development, and ultimately would only serves to fuel the argument to do away with all public tables.

This is why I challenged members, rhetorically, to look at the ranks and decide for themselves if it is doing a good job. I believe it is and I hope that most of you agree.

Having said all that, we are currently discussing a small tweak to the system to handle the "gatekeeper" problem. That is, one particular member holds the current minimum requirements to enter the first viewable table (cubes). If they are surpassed, that member will become the new gatekeeper and if they pass others they hand on the baton to another in that table. Consequently, the lower limit will begin to slowly rise. The balance against that is that the size of the problem set also rises meaning there are more problems to pick from, but this is somewhat contrary to one of our goals. So a possible solution is to only show members who have solved a problem within a specific time frame, thereby temporarily removing long term inactive members; of course, they would immediately reappear if they returned and started solving problems again.

[thedoctar]

There's a really interesting youtube talk by the creator of Lichess, an online chess site [link]https://www.youtube.com/watch?v=LZgyVadkgmI[/link]

He discusses the prevalence of cheating in online chess, (shadow)banning cheaters and what cheaters say when they are caught (plus the lies, intimidation, threats that follow).

Maybe PE team has seen it or has experienced similar things. Maybe they might learn something from the talk.

But it's interesting from a “user” perspective. I don't have to deal with all that social crap

Which is just to say that dealing with cheaters is also a problem, in fact far worse, in online chess; particularly because chess engines are better than humans.

[hacatu]

Yeah, even though I disagree with the recent updates, I can't overstate how much I appreciate all the work the whole development team does! At the end of the day, the problems speak for themselves