Simply rating new accounts at 900 may not be a good solution. This would open new ladder abuse possibilities, as it would be easier to get a large amount of ladder points and gold by making new accounts. For example, a player rated at 1350 might be stuck at that rating and losing as many ladder points as winning every week, thus not getting much gold. Opening a new account would not benefit him / her much at all in the current system, as he would start at 1300 with a large uncertainty. But he would probably have easy time in 900-1200's similar, getting a lot of gold for many weeks in row.

The starting rating must be as good an estimate of 'the average rating' as possible, since there is no way to know whether a new account belongs to a new or an old player. Then again, if the system was suddenly changed to give new accounts a rating of 900-1000, that would simply lead to the average rating converging to that number (assuming there will always be new players who actually play ranked games). This is because when someone's rating goes up, someone else's has to go down.

Few comments:
- Regarding ladder abuse… I get it. Nobody likes cheaters. This is a great game and it sucks that people will attempt to abuse a good thing.
But please, keep some perspective. You find cheaters in all aspects of life. I’ll reserve my outrage for the abuses that do not occur in the context of an on-line game.
The winner of the ladder gets 120 g. That is $2.99 US if buying from the website. I don’t think anyone is planning their retirement around gaming the ladder system in TFW.
IMO, though, right now it is more important to get and keep people playing the game. Once a sizable player base is established, concerns like this should (and I hope will be) more significant. Also, abuses due to second accounts are already something the developers have effectively limited. Under the new point system released a few weeks ago it is significantly harder to game the ladder. Assuming I read the formula in Jed’s post a few weeks ago correctly, here are the ladder points which would result from not-so-randomly chosen pairings. All numbers are assuming this is the first game between the opponents.

R-w 1300 U-w 350
Game ? R-l U-l Points
YES 900 75 -8
YES 1000 75 -2
YES 1100 75 4
YES 1200 75 9
YES 1300 75 15
YES 1400 75 21

R-w 1300 U-w 250
Game ? R-l U-l Points
YES 900 75 -14
YES 1000 75 -4
YES 1100 75 6
YES 1200 75 16
YES 1300 75 26
YES 1400 75 36

R-w 1300 U-w 150
Game ? R-l U-l Points
NO 900 75 -29
YES 1000 75 -8
YES 1100 75 13
YES 1200 75 33
YES 1300 75 54
YES 1400 75 75

R-w 1300 U-w 50
Game ? R-l U-l Points
NO 900 75 -93
NO 1000 75 -26
NO 1100 75 41
YES 1200 75 109
YES 1300 75 176
YES 1400 75 243

The Game ? indicates if TFW would allow a similar match between players with the indicated rating and uncertainties. If no, the ladder points are purely hypothetical.
First, please note that players with high uncertainties get far fewer ladder points than players with low uncertainties. Second, please note that players with high rating can actually get negative points for beating up lowly ranked players. Makes me wonder if I did the math right, but it was checked several times. Maybe Jed can let us know if I did it correctly.
If someone wants to cheat, I say let them try. Their 2nd account, when new and of high uncertainty, will get them little ladder points if they win. If they win, all that happens is they play better opponents. If a cheater can beat players above 1350 regularly enough to win the ladder, then he/she is an idiot for cheating when there is no need. If they lose the first few games, the whole point of creating a 2nd account is lost. And remember, these numbers are for 1st games against a given opponent. 2nd and 3rd games will get even less points.
Finally, if the suggestions originally put forth by Zavia are adopted, ladder abuse becomes impossible due to the 7 day waiting period. But, honestly, even now, the potential of ladder abuse is highly over-rated.
- Even experienced but not expert players like myself crush 1300 (350) players. And like Yaron said, it is no fun for either party.
- DJ… Drinking Newcastle tonight. Good stuff. My, this makes me sound like a drunk, but honestly I only have 2-3 beers a week. With alcohol, quality over quantity is the way to go.

wern't we suppose to knock our heads togather and get something going?
Altren didnt get back to me when I asked for his philosophies and all dat...


Depends on the exchange rate...


Yes, been taking quite a few crushings lately...

... Why do i keep seeing my name around? I havn't been posting....

Also, bz wit midterms, cant really afford to read all dat. May mix my studies up. Haha.

Hmm I think we should put up suggestion threads on diffrent aspects of tfw.

Ratings is among the things TFW REALLY needs to improve.

Two things should be a factor in Rating:
Total Wins against Players.
Win Loss Ratio

Then smaller modifiers can be:
Age of account
Tournaments and what not give minor "bonuses" to your rating
and what not.

Example:
[Total Wins*]x([Win/Loss])+minor bonuses

Total wins being againts a human player, and is a 1v1. Sims and challenges are included.
Win/Loss above will actually screw with the ratings rather heavily. A better formula for win/loss is (([Win/Loss]-1)/3)+1. Took me a while to find a way to bring it closer to "1".

So anyway, say you won an EC. Winning it gives you a whole 6 to your rating for example.

I agree that losing 1300 rating so rapidly is frustrating for new players, they should start at 900.
AND
please, make MORE ladders. 1 for "elite", 1300+ rating, 1 for others and 1 for newbies. With appropriate prizes there would be NO abuse

I will share what I have seen in other games. Give ranks or "honor badges" for players. For various achievments.
for example>
Baby>child>young>....>ancient: based on the age of the account
Soldier>Warrior>Leutenant>.....General: based on win count against players
Trickster>Illusionist>Sorceror>Mage>Arch Mage: based on positive Win/loss ratio against players
ocassional>regular>usual>ADDICT: based on number of games played or time spent on line
peasant>noble>lord>prince>king>emperor: based on real money spent on the game
novice>apprentice>expert>master>grandmaster: based on number of won tournaments.
and so on....
This would be a colorful addition and would represent the players more exactly than his rating or anything else. After, you could make some events for XX class of players (or above).

Once upon a time I thought limitations based on rating was good idea. But now that I don't play often enough to keep my uncertainty down and maintain a high rating, I think it's a terrible idea. Rating attrition, though necessary in some regards, has made the rating system a bit deceiving.

For example, my constructed rating is currently at 1300, and limited 1500, both with relatively high uncertainty. But less than a year ago, both of my ratings were over 1700. Furthermore, I haven't won an absurd number of games, so there's really no reliable way for a new-ish player to have any idea what they're getting into. And the problem is compounded by skilled players who play under new accounts.

In some ways I think the rating system would be more useful if it worked like chess. Uncertainty has, imho, stripped the rating system of most of its "statistical meaning". Perhaps a better implementation would have been for players to be knocked off the leader board for lack of activity rather than rating attrition due to the slow upward march of uncertainty.

got a link that better explains it?

we already have that, its next to your name.

In a nutshell, you have a sort of "uncertainty" until you've played enough games to get a 'real' rating. From then on out, your rating increases/decreases purely based on the ratings of players you beat/lose to... and there's no persistent accumulating uncertainty to cause the steady rating attrition we see in TFW.

US Chess Federation Rating System described in detail here:

http://math.bu.edu/people/mg/ratings/rating.system.pdf

One of the reasons uncertainty was added was so that players who didn't play often or who stopped playing would eventually fall off the leader board. I think they should fall off from lack of activity, not from their uncertainty level rising to n00b levels.

This is a minor issue compared to developing a larger gaming community, but it's something that would be good to fix someday. It would give rating "meaning" again.

I have a suggestion.

When i first started out recently there was some confusion over the whole game, i didn't know why i couldn't get something to cast, and i used to play mtg, the tutorial, though helpful is not clear enough in certain aspects. I often find players who never played mtg, but want to play this one, and cannot understand how to use their cards, or they build in a corner, or right outside of their area of control but w/ no purpose as to why they are. we definately need a new tutuorial and then an advanced tutorial.

Next a campain, as suggested in diff thread would help keep new players who can't seem to find a human to beat...it gets discouraging when you cant beat anyone at all, it makes you feel stupid.