What were the changes? Can you give us a brief summary of what we should expect the AI bots to do differently now?
Also, thanks for posting here! I really like the program, and it's cool to be able to discuss issues with someone who's willing to change it for the better.
First round of changes basically made bots avoid particular colors during successive draft rounds until the last round where they're just picking two colors (when they can). They now also tend to grab the 'best' card more strictly (it used to use a lot of fuzzy probability)
Private Mod Note
():
Rollback Post to RevisionRollBack
--------------------------------------------------------- TappedOut.net -- A decently not bad mtg deck builder
I've had my cube up there for a while now and I recently started having troubles with it....
I'm eager to test with the "new and improved" bots but it looks like I'm currently unable.
link?
Private Mod Note
():
Rollback Post to RevisionRollBack
--------------------------------------------------------- TappedOut.net -- A decently not bad mtg deck builder
When I see Healing Salve, I'm often like "Oh girl, I wish I could turn every card into this." Thanks they removed the gain life part, otherwise this would have been broken.
Not only am I still getting terrible bot picks, but it's running slower and more sluggish right now than ever. Every single pick winds up with a "draft error; reload browser" ...in both Firefox and Safari.
Cant even get through a draft.
edit: and it just froze between packs 1 and 2 two drafts in a row. ...and I picked up a P1P7 Ancestral Recall...
Not only am I still getting terrible bot picks, but it's running slower and more sluggish right now than ever. Every single pick winds up with a "draft error; reload browser" ...in both Firefox and Safari.
Cant even get through a draft.
i am having this problem, too. too much cube drafters for the site to handle?
When I see Healing Salve, I'm often like "Oh girl, I wish I could turn every card into this." Thanks they removed the gain life part, otherwise this would have been broken.
It may be important to note I am also experiencing great load times on Chrome.
Also when I first uploaded my cube I got an Ornithopter error. I thought maybe the problem lied within how long the title of my cube was so I renamed it to a shorter name, and it worked. Not sure if it was a fluke but thought I'd mention it for others to try.
Not only am I still getting terrible bot picks, but it's running slower and more sluggish right now than ever. Every single pick winds up with a "draft error; reload browser" ...in both Firefox and Safari.
Cant even get through a draft.
edit: and it just froze between packs 1 and 2 two drafts in a row. ...and I picked up a P1P7 Ancestral Recall...
Private Mod Note
():
Rollback Post to RevisionRollBack
--------------------------------------------------------- TappedOut.net -- A decently not bad mtg deck builder
Is there a way to isolate the bot pics based on how highly they're picked from exclusively cube drafts?
That way, the bot AI would get better the more mock cube drafts are done, and it would ignore the data from the corresponding limited environment.
This would stop P1P1 bombs from Scars (like Myr Battlesphere) from being picked higher than cards you don't have draft data for (like Library of Alexandria).
Making AI for a cube is going to be difficult because picking an archetype is pretty difficult to code. For instance, Myr Battlesphere might be P2P1 if you've got recurring nightmare and/or survival of the fittest, whereas if you're playing counter-burn he's kinda useless. Akroma is great in a reanimator deck, but not so much in other decks that could potentially use her. My initial thought would be to rank cards based on how high humans draft them, but how good an individual card is varies more on what strategy you're going for much more than in a normal draft.
When I see Healing Salve, I'm often like "Oh girl, I wish I could turn every card into this." Thanks they removed the gain life part, otherwise this would have been broken.
Now after I add the basic land to my deck, it tells me that my EDH general's name is spelled wrong, even though the "booster draft deck" button is selected.
Now after I add the basic land to my deck, it tells me that my EDH general's name is spelled wrong, even though the "booster draft deck" button is selected.
unrelated temporary error. =)
Private Mod Note
():
Rollback Post to RevisionRollBack
--------------------------------------------------------- TappedOut.net -- A decently not bad mtg deck builder
Is there a way to isolate the bot pics based on how highly they're picked from exclusively cube drafts?
That way, the bot AI would get better the more mock cube drafts are done, and it would ignore the data from the corresponding limited environment.
This would stop P1P1 bombs from Scars (like Myr Battlesphere) from being picked higher than cards you don't have draft data for (like Library of Alexandria).
Sounds like the only way to do it, really. I'm gonna work on this.
Private Mod Note
():
Rollback Post to RevisionRollBack
--------------------------------------------------------- TappedOut.net -- A decently not bad mtg deck builder
Yeah, I did a draft a bit ago and one of the bots took P1P1 Vivid Marsh. Another took P1P1 GAAIV.
MTGS Average Peasant Cube 2023 Edition
Follow me. I tweet.
What are your expectations there?
If you look at the cards on TappedOut, you can see the AVG draft picks. I don't really see them as being hugely different.
TappedOut.net -- A decently not bad mtg deck builder
Defintely thanks for mentioning this. I've had a feature freeze for a while dedicated simply to making my load time superior. Glad someone notices. =)
TappedOut.net -- A decently not bad mtg deck builder
First round of changes basically made bots avoid particular colors during successive draft rounds until the last round where they're just picking two colors (when they can). They now also tend to grab the 'best' card more strictly (it used to use a lot of fuzzy probability)
TappedOut.net -- A decently not bad mtg deck builder
link?
TappedOut.net -- A decently not bad mtg deck builder
i can understand the gaaiv pick. maybe that bot just really loves blue/white control the way you'd first pick carnophage because you love black aggro.
but the vivid pick is hilarious. there's no way to explain that one away.
Cant even get through a draft.
edit: and it just froze between packs 1 and 2 two drafts in a row. ...and I picked up a P1P7 Ancestral Recall...
My 630 Card Powered Cube
My Article - "Cube Design Philosophy"
My Article - "Mana Short: A study in limited resource management."
My 50th Set (P)review - Discusses my top 20 Cube cards from OTJ!
i am having this problem, too. too much cube drafters for the site to handle?
It may be important to note I am also experiencing great load times on Chrome.
Also when I first uploaded my cube I got an Ornithopter error. I thought maybe the problem lied within how long the title of my cube was so I renamed it to a shorter name, and it worked. Not sure if it was a fluke but thought I'd mention it for others to try.
TappedOut.net -- A decently not bad mtg deck builder
Still have uber-bombs going WAY late, but that's cool. It's a fun tool... it can't be perfect.
My 630 Card Powered Cube
My Article - "Cube Design Philosophy"
My Article - "Mana Short: A study in limited resource management."
My 50th Set (P)review - Discusses my top 20 Cube cards from OTJ!
For instance, take Jace.
Average pick without cubes = 1.2
Average pick with cubes = 3.0
Any suggestions?
TappedOut.net -- A decently not bad mtg deck builder
That way, the bot AI would get better the more mock cube drafts are done, and it would ignore the data from the corresponding limited environment.
This would stop P1P1 bombs from Scars (like Myr Battlesphere) from being picked higher than cards you don't have draft data for (like Library of Alexandria).
My 630 Card Powered Cube
My Article - "Cube Design Philosophy"
My Article - "Mana Short: A study in limited resource management."
My 50th Set (P)review - Discusses my top 20 Cube cards from OTJ!
My 630 Card Powered Cube
My Article - "Cube Design Philosophy"
My Article - "Mana Short: A study in limited resource management."
My 50th Set (P)review - Discusses my top 20 Cube cards from OTJ!
unrelated temporary error. =)
TappedOut.net -- A decently not bad mtg deck builder
My 630 Card Powered Cube
My Article - "Cube Design Philosophy"
My Article - "Mana Short: A study in limited resource management."
My 50th Set (P)review - Discusses my top 20 Cube cards from OTJ!
Sounds like the only way to do it, really. I'm gonna work on this.
TappedOut.net -- A decently not bad mtg deck builder
TappedOut.net -- A decently not bad mtg deck builder
edit: it may just be not enough people have drafted yet but, avalanche riders doesn't seem to have an average pick.
2: nor does rolling earthquake.
If he switched the picks over to cube picks only, there may not be any data for them yet. So draft them where you think they should go!
My 630 Card Powered Cube
My Article - "Cube Design Philosophy"
My Article - "Mana Short: A study in limited resource management."
My 50th Set (P)review - Discusses my top 20 Cube cards from OTJ!