William Stein: some panel discussion statements

# What are the strength and weaknesses of open source versus closed source software?

• Can we trust results of computer algebra systems/ mathematics
  • software that is closed source? Or of any software at all?

No. No. We can't trust anything automatically -- there are only levels of trust. I'm more interested in usefulness. Extendability. Etc.

• Should a paper submitted for publication in a journal be
  • rejected if it fundamentally depends on closed source software? Or on any software at all?

It depends on the quality of the journal. For a top quality journal, I have rejected such papers before and will again. Depending on software "at all" -- that's just a fact of life.

# Is it possible to provide a truly viable open source alternative to

• Maple, Matlab, Mathematica, and Magma that will survive the standard 30 year software lifespan and be well supported? No, it's completely impossible, since it's so frickin' hard. I don't care, I will do it anyways.

# There have in the past been many open computer algebra projects,

• which start out with much enthusiasm and excitement, but every single one has since failed. Why should Sage be any different?
  • It is different than those projects in many ways:
    1. GPL
    2. Not at all "design by committee" -- everything is driven by immediate needs
    3. Do not reinvent the wheel
    4. Large number of developers
    5. Already SIX Workshops from Feb 2006 to Nov 2007, and many more to come.
    6. Me -- I want this to succeed very very much.

# Is it likely that any of the following mathematical software systems

• will be open sourced within the next five years: Fermat, Kash, Matlab, Maple, Mathematica, Magma, Mupad, Reduce?

Fermat or Reduce might... they are both by "out of touch" almost retired mathematicians... but who cares?

Or that they will be replaced by viable OSS alternatives?

I hope so.

# Do 'we' have to protect our OSS using a copyright such as the GPL,

• or should we just release using a BSD-like license to avoid the hassle? [Discussion on this point will be strictly limited.]

We have to use the GPL, because PARI, Singular, etc., already did. There's really nothing more to say.

# Discuss ideas for funding the creation of open source mathematical

• software, and of improving the *quality* of open mathematical software (e.g., testing, bug days, trac, etc., language choices, etc.).

• Funding:
  • workshops cost a lot, so get them funded by institutes...
  • hire undergrads "student/faculty research".
  • creative NSF grants (e.g., the new CDI initiative) -- I will
    • likely be on at least two CDI's because people want to "use" Sage to pump up their apps.
  • donations (tax-free)

# Should software implementations be treated more like journal

• submissions, being refereed and hence putting computational work on the same footing as more theoretical work in the context of academic credit? (cf JSage)
  • In some cases this would be very valuable for the mathematical community.
  • It is also greatly improves code quality.
  • Axiom: Non-refereed code is broken.

# Could (or should) Sage be packaged and distributed as a linux

• package in one (or more) standard formats (.deb, .rpm)?
  • Yes, it definitely can and should be. Give me money to hire
    • somebody to do this. Or maybe Michael Abshoff will. Definitely. I'm very surprised it hasn't happened already.