What should basic_strategies be?

[drvinceknight]

On #296 (which has now been merged in) @langner, @marcharper and I had a chat about basic_strategies. Marc wrote a nice function that defined basic strategies however this changed the basic_strategies to:

Alternator
Anti Tit For Tat
Bully
Cooperator
Defector
Soft Go By Majority
Suspicious Tit For Tat
Tit For Tat
Win-Stay Lose-Shift

which Karol and I thought (please correct me if I'm speaking incorrectly on your behalf Karol) was a bit too much for the purposes of a demo - which is what I thought basic_strategies was for.

I wonder however if a demo would be better served by:

strategies = [axelrod.Alternator(), axelrod.Cooperator(), axelrod.Defector(), axelrod.Random(), axelrod.Tit For Tat()]

As this shows how to access strategies (using a JuPyTer notebook, this would be quite nice with tab completion for example), then basic_strategies can be used to show a bit more of a complicated tournament?

What do people think? I think it's a good one to chat about :)

Note that the actual work for this has more or less already been done and can be cherry-picked.

[meatballs]

Personally, I think it's useful to have a dynamic generation of strategies which are 'basic' as defined by their classifiers.

However, if people also find it useful to have a static list for demo_strategies then I have no objection to that.

[drvinceknight]

Yeah I think I'm coming around to that also. Whether or not you use
demo_strategies is up to the particular demo... I think I might not
actually use it myself and just use tab completion to show grabbing each
strategy...

As new dimensions come in, we could also potentially include them in the
is_basic function (just thinking put loud, not very relevant right now).

On Wed, 16 Sep 2015 08:15 Owen Campbell notifications@github.com wrote:

Personally, I think it's useful to have a dynamic generation of strategies
which are 'basic' as defined by their classifiers.

However, if people also find it useful to have a static list for
demo_strategies then I have no objection to that.

—
Reply to this email directly or view it on GitHub
#297 (comment)
.

[langner]

I'm fine with the dynamic one, too.
On Sep 16, 2015 12:24 AM, "Vince Knight" notifications@github.com wrote:

Yeah I think I'm coming around to that also. Whether or not you use
demo_strategies is up to the particular demo... I think I might not
actually use it myself and just use tab completion to show grabbing each
strategy...

As new dimensions come in, we could also potentially include them in the
is_basic function (just thinking put loud, not very relevant right now).

On Wed, 16 Sep 2015 08:15 Owen Campbell notifications@github.com wrote:

Personally, I think it's useful to have a dynamic generation of
strategies
which are 'basic' as defined by their classifiers.

However, if people also find it useful to have a static list for
demo_strategies then I have no objection to that.

—
Reply to this email directly or view it on GitHub
<
#297 (comment)

.

—
Reply to this email directly or view it on GitHub
#297 (comment)
.

[marcharper]

If you do want to get the basic_strategies code back, just rollback the last commit (c031c23), which removed is_basic and the associated tests. Then just put this back in for demo_strategies (if you want this in the library):

# Defined by fiat for demo purposes
demo_strategies = [Alternator, Cooperator, Defector, Random, TitForTat]

[drvinceknight]

I'm on it :)

[drvinceknight]

I believe #298 takes care of this now :)

Sorry for being a pain @marcharper.

[marcharper]

No worries