Info Vilesilencer Directory Owners Forum and Free Directory List » The New IVS Site » IVS Ranking System & The Top 100

[Dan]

The Top 100 rankings have been busted for a bit.

Basically the numbers don't reassign, even though directories will move up and be in valid positions.

e.g. The top 20 is - #1-17 are fine, then #18 is missing, then it goes #19-21

What should happen is that #18 should be reassigned, such that #21 which is AMRAY would actually be #20 because it is in the #20 slot.

That's the first thing, THEN the actual calculations themselves need to be fixed. I don't have any plans to change the way the rankings are based on this update... it's based primarily on age (where age = time on the IVS list). PR and Alexa are used to tie-break and the 4th component is the votes system. The more votes the more points towards the rank, though this is not a huge factor to prevent gaming.

The way the age works, it actually does max out at some point. Such that the directories that have been on the longest once they've received their maximum value, will get caught by other directories whose age has slowly built over time. What then happens is the alexa/PR/votes takes over to tie break, so older directories could move downwards.

e.g. I would expect Pegasus to move a few places downwards, as despite being on the list for some time, it's PR has dropped significantly. I expect this is due to the fact that it appears to be under a penalty.

2.0 Websites is down to PR0, and there are a few other PR drops from sites that were 5 or 4, so these could significantly shake up the top 20, and then below that I would expect some more to come through.

There are no plans to add the ranking/review system to the calculations, though this could happen in future and tie in with the votes just to round that whole section out.

I have pondered on including a new editorial mentions factor. Where the number of authority type resources in the field that also mention the directory would factor into the rankings. It's likely to happen, as it ties in with a blog post I am doing that contains lots of detailed research, so stay tuned for that.

That would be a manual factor, so I may only include it as part of the top 100.

There are other factors I have pondered on including too. Some are automated some are manual. The manual ones I think its difficult to get too broad with. If we start applying that to every directory we lose scope and it becomes an exercise in futility because we never actually get all the directories in the list done. If it is for the Top 100, and maybe the General Paid, to rank those 2 lists, then it would be worthwhile.

All of these factors in the end could just come down to one IVS score (an actual score that we would display or not-display but just use to assign the rank).

Since it's much more likely that these scores would be unique we could drop the PR/Alexa/Votes emphasis off almost completely, and have far more editorial methods with which to rank the directories.

To be fair I doubt it would create a huge difference in the lists. The good directories are always going to rank near the top. There might be a few hidden gems that we've missed or a few slightly dodge ones that rank higher than they should, but all in all the core of the list is solid. Always has been.

Probably leave that till it's fixed, as that is first priority.

So over the next week or so the rankings could be all over the place. Just be aware we are on top of it and having it sorted to...

[Dan]

As I expected the rankings just went haywire. They've re-sorted themselves in some weird manner.

There's 0's showing for rankings, these are being ranked above number 1, so basically any newer directory is now in that slot.

It'll be like this till the rankings are fixed so just bear with us.

[Dan]

I've manually moved the 9 directories showing 0 by giving them sequential ranks, so they return to page 2. The list should have re-structured itself now.

The rankings will still be incorrect but the placements are what they were before.

[mimika]

Yep, it looks like it's working now... finally

Talking about '0's.
Why do you need metrics from bottom Up?
I can bet no one is looking forward to find which directories have least PR or Alexa but with your system they will get these metrics if they accidentally push wrong button.

BTW, is there a way we can edit old description?

mimika

[Dan]

It's not working. I haven't had it fixed yet. It's still broken.

Metrics from bottom up?

Do you mean: http://info.vilesilencer.com/top/order_by/google_pr/up

I may look at taking out the [order the list by metric] arrows that would order it by PR0 or by Alexa 5 million. I doubt anyone even uses them... but yeah if they wanted to order by metrics, then I guess pushing the wrong button would have them looking at directories the may not be interested in.

How do you edit old description? Well since I wrote it you can't. But I can. Are you asking me to re-review the directory and write a new description for it?

[Dan]

Ok. The ranking system is now fixed.

As I stated before I may apply some new manual and auto factors to syphon/hone the list more. But it could take a while to implement these as the specifics need to be worked out.

If anyone sees any anomalies (there's a directory in 0 position that I am aware of just now) do let me know.

Also because we had a mini-glitch in the pr/alexa system *some* of those mightn't have updated (you'll see some with a 0 Alexa score). Once these are re-run through the system they should be ok, and then the calculation will take effect and change again.

[Dan]

Having run the actual calculation script for myself manually yesterday, I can see that this isnt fixed.

It works. The calculations are done correctly, and the results that we get ARE correct.

However, rather than just ranking the Top 100, the system is checking against every URL in the system (something the original coder has made a huge misunderstanding with --- since I never wanted that done in the first place).

The system was supposed to skip every other directory type and all directories listed as being in quarantine, and only rank the Top 100 directories, where those directories are displayed online in the system... That would literally be 110-120 URLs versus the several thousand that are now being checked.

The check is quick, but that huge amount of server overhead is what causes the crashes, especially when the monthly recalculation was done and compared. Because that then became a doubling of the several thousand calculations and always caused the server to malfunction.

I'm going to have the new coder look into changing what we have to cut this server overhead and just create a nice smooth stripped down ranking system that is fast. It will also allow me to input more values into the system to get more consistent and accurate ranks. Though what we have now is pretty close!