Project Euler Forum

Aha! I think I figured out why the score tables were showing strange data before. It related to non-registered users trying to view them. For the moment, at least, I've sorted that problem.

Taifu, it was more of an issue previously when accessing the score tables required a query on the database to be performed each time and I didn't want search engines to recursively follow links, hogging bandwidth.

I know that registering isn't exactly a major chore, but I suppose that the real question is whether or not members would be happy with score tables being made "public". And would I have to "serve" different score tables for visitors/logged-in members? For example, although the "profile" link would be disabled, the score tables contain "additional information" and whether or not that member can be contacted.

euler wrote: I know that registering isn't exactly a major chore, but I suppose that the real question is whether or not members would be happy with score tables being made "public". And would I have to "serve" different score tables for visitors/logged-in members? For example, although the "profile" link would be disabled, the score tables contain "additional information" and whether or not that member can be contacted.

For my part, I have no problem with visitors having access to the same information about me as registered users.

euler wrote:Taifu, it was more of an issue previously when accessing the score tables required a query on the database to be performed each time and I didn't want search engines to recursively follow links, hogging bandwidth.
I know that registering isn't exactly a major chore, but I suppose that the real question is whether or not members would be happy with score tables being made "public". And would I have to "serve" different score tables for visitors/logged-in members? For example, although the "profile" link would be disabled, the score tables contain "additional information" and whether or not that member can be contacted.

Colin,
I understand your concern, but I think that with good caching it's possible to let scores table become public.
Moreover I think 99% (at least) of members would be proud to show their new level scores table to relatives and friends
Ciao.
Marco aka Taifu.

Taifu wrote:Moreover I think 99% (at least) of members would be proud to show their new level scores table to relatives and friends

That may be true, but the remaining 1% should have an option to "opt out", so to speak.

elendiastarman wrote:

Taifu wrote:Moreover I think 99% (at least) of members would be proud to show their new level scores table to relatives and friends

That may be true, but the remaining 1% should have an option to "opt out", so to speak.

Level scores are available to every member and everybody could be a member. I'd understand your observation if Project Eulero were a reserved club.

Btw I don't want to push this because I was just nitpicking: I do *love* this site

Ciao.
Marco aka Taifu.

Is it possible to have a section on Project Euler, like the problems section, but only with the 25 recent problems (either separate from or added onto the problems section)? This would help a bit with functionality -- if people wants to improve their current performance, they're only going to want to see the last 25 problems. They can easily discern recent problems in the current setup only if the problems are sorted by ID, but not if by sorted by difficulty.

Or we can highlight/color these problems differently?

tklalmighty wrote:Is it possible to have a section on Project Euler, like the problems section, but only with the 25 recent problems (either separate from or added onto the problems section)?

Why can't one just select "Sort problems by ID in descending order" from the problems page and just look at the top few?

Thumbo wrote:

tklalmighty wrote:Is it possible to have a section on Project Euler, like the problems section, but only with the 25 recent problems (either separate from or added onto the problems section)?

Why can't one just select "Sort problems by ID in descending order" from the problems page and just look at the top few?

Yeah, that's what I meant by "added onto the problems section". That's probably better than having a completely different section. (I still think the scores section and the stats section could be merged somehow, but that's a different problem.)

Edit: I can't read ...That is probably the easiest way to do that within the current set-up, but if I wanted to, say, sort problems by difficulty?

Actually, speaking of that section, I have a minor nitpick. When I have my problems sorted ID ascending, the actions show something like "Sort problems by difficulty in descending order." which is a little confusing. I know it's obvious from looking at the output below what ordering you use, but it doesn't explicitly say it anywhere. I think maybe dropboxes would be better. That way we can show "Displaying [solved/unsolved] problems sorted by [ID/difficulty] [ascending/descending]." where each [] is a dropbox. That might be a tad bit clearer.

OK, so I know it's been over a month since the last post here, but I'm passing along a request from my girlfriend that the "n% genius" text be restored along with the green bar on the profile and problems pages. She says that telling her friends I'm 100% genius is much more satisfying than telling them I'm an icosahedron.

--SMQ

I have a question.

Till now I've only solved 126 problems, but I've finished all the last 25 problems.
I thought when I solved all the last 25 problem, my rank won't be changed in the "Top Eulerians" until a new problem is added.

However, my rank keeps slipping. It seems when others finished the last 25 problems, they will get ahead of me.
Does that means the ranking is sorted by both the performance and the solved number, or maybe the time of solving the last problem?

I have no inside knowledge of the internals of the PE system, so I'm just speculating.

It looks to me like the order of the top Eulerians is determined by:
a. Current performance (how many of the last 25 problems you've solved)
but if those are equal, it order is determined by
b. Total number of problems solved
and if those are equal too, order is determined by
c. When you last solved any problem.

Here b and c form the normal ranking procedure.

I'll assume you would like the order to be determined something like this:
a. Current performance (how many of the last 25 problems you've solved)
but if those are equal, it order is determined by
b. When you last solved one of the last 25 problems.

There is something to be said for both schemes, but I think it will be tricky to do the latter because there would need to be a timestamp stored for all of the last 25 problems, rather than 1 timestamp per participant, so that would mean a lot of programming work for someone.

Maybe what could be done is:
a. Current performance (how many of the last 25 problems you've solved)
but if those are equal, it order is determined by
b. When you last solved any problem (ETA: in reverse order)

This way there would probably not need to be many changes to the system, and the Current Performance would become slightly better measure of recent activity. The Current Performance will still be affected by problems not in the last 25, but you would have more influence over it.
I have no idea though if this is a feasible proposal.

Edited to add:
Correction: The first order I listed is not how it is done now. What is done now is the CP+time, with oldest timestamp at the top. My third proposal was intended to be with the newest timestamp at the top.

Yes, this is what I like

a. Current performance (how many of the last 25 problems you've solved)
but if those are equal, it order is determined by
b. When you last solved one of the last 25 problems.

But I think this is the current way it's sorted

a. Current performance (how many of the last 25 problems you've solved)
but if those are equal, it order is determined by
b. When you last solved any problem.

The weird thing is that if I keep solving other problems, my rank would keep slipping.
However, if I don't solve anymore problems, my rank wouldn't change.

Maybe the first scheme you mentioned is more reasonable.

I corrected my previous post - I misunderstood the current situation.

I agree, the order of those with equal CP should be done either by:
A) first person to reach that CP (i.e. longest time since one of the last 25 problems solved)
B) Normal rank order
or
C) last person to solve any problem (i.e. shortest time since any problem solved)

Definitely not the way it is now.
(A) may be hard to achieve. (B) or (C) should be fairly easy.
Option (C) is probably a bit unfair in the group with CP=25, but reasonable for lower CP values since it is a measure of activity.

A nice piece of reverse engineering; you've correctly determined the "internals of the PE system".

It would require more work adding an additional timestamp for the most recent 25 problems, but I am happy to do it. It would also mean a few temporary "glitches" in the ranking of Eulerians as the database field would be empty for members until they solve one of the recent problems, but it would eventually settle down.

In the meantime I've changed it to rank Eulerians on: how many of recent 25 solved, how many solved in total, last solved timestamp.

In some ways this is fair, because those that have solved more problems should get recognition. But equally they receive recognition in the other tables and the whole point of the Eulerians table is to allow those who work hard on the most recent problems to receive recognition.

Let me know which is the preferred option.

euler wrote:In some ways this is fair, because those that have solved more problems should get recognition. But equally they receive recognition in the other tables and the whole point of the Eulerians table is to allow those who work hard on the most recent problems to receive recognition.

Let me know which is the preferred option.

I certainly prefer the second option (ignore the number of total solutions) as the Eulerians table should only be based on the current performance.

I've made the changes. It will take time to start functioning properly because the "recentsolved" timestamp field will be empty for all members until they solve a "recent" problem.

It now sorts Eulerians on: (i) how many recent problems solved, (ii) timestamp of most recent problem, (iii) how many problems solved in total, (iiii) timestamp of last solved.

If the "recentsolved" field is empty then it temporarily sets it to current time before it sorts which means that members will be ranked on (i) then (iii). However, if a member now solves a recent problem the timestamp will beat current time so they will go to top of that tier. This slightly undesirable outcome will settle down after members begin solving recent problems. It is necessary because if the "recentsolved" field was left blank then when a member solves a recent problem they would be below all other members in the new tier that haven't recently solved a problem, regardless of having solved more problems. At least this way it temporarily rewards those that solve a recent problem.

Please keep an eye on it and let me know if you notice any bugs.

I think the new system is fine except for one thing: as there are only five Platonic solids, does that mean that the problems will stop at 249? That is not far away...

On your 250^th solved problem, you become a sphere (or is it a solid ball?), that'll last until problem 10,000 or so

rmillika wrote:... as there are only five Platonic solids...

Actually the team have been keeping quiet about this, but your question has forced a response. Although some time off yet we're working on a radical new proof surrounding the discovery of a sixth Platonic solid. Obviously we don't want to rush it, but we were planning to go public with it at the beginning of April next year.