Re: [HoTT] How to make software without money

[Timothy Carstens <intov...@gmail.com>]

[-- Attachment #1: Type: text/plain, Size: 6766 bytes --]

In the computer business, technologies which solve essential problems tend
to survive; extensible technologies which solve essential problems tend to
grow.

When you look at the largest open source projects, the majority of the
actual code tends to come from people working in large companies that
depend on the tech stack. Linux, for example, has benefited enormously from
IBM's investment of engineering-hours into the project. Ruby on Rails is
another example, as are many other web-related frameworks.

The SAGE math environment was a noble effort to displace MAGMA and similar
systems, which were benefiting from code contributions from academia, but
which remained commercial and not widely available. Later SAGE would evolve
to compete with Mathematica and other math suites. Unfortunately, the
essential problem being solved was a matter of taste-in-licensing, which is
not always essential-enough for a project to take hold.

Coq, Agda, Lean, etc, all solve an essential problem, namely that of
providing a theory of higher-order machine-checked proofs. The
implementations tend to be extensible, at least for experts. Different
segments of the CS and math community are becoming aware of these tools and
producing new, impressive artifacts all the time. Overall it's a small
community, but it's growing. At the very least, the diversity of artifacts
demonstrates that these tools are expressive-enough to be useful in many
domains, not just in principle but in practice.

Though it is currently difficult to apply these methods to problems of
interest in industry, progress in the field of program logics is real and
has lead to real tools (Infer, at least) with real users. The ability to
prove properties of programs inevitably leads to the desire to prove
properties of fairly technical programs, and in that way a bridge to more
abstract, less-computery branches of math will emerge. An ecosystem will
form, and pure research which contributes to this ecosystem will begin to
take on its own appeal. Other paths to industry-awareness may exist, this
is just the one I've been tracking.

In my opinion, patience will be rewarded.

-t



On Sun, Jun 12, 2016 at 4:05 AM, Andrej Bauer <andrej...@andrej.com>
wrote:

> While the efforts expanded on Coq and Agda are truly impressive and
> chivalrous, both pieces of software compare poorly to something like
> Mathematica in terms of software quality (documentation, professional
> GUI design, technical support, cloud support, etc.) This is not a
> criticism of the Coq and Agda teams, just an observation which in
> Slovene could be summarized by the phrase "that's the music you get
> for the money you paid".
>
> Also, I am not trying to start a war with the lurking Knights of the
> Open Source. I am just saying we have no idea how to bring open-source
> mathematical software to the level of professional software without
> sacrificing the careers of several PhDs and at least one tenured
> professor.
>
> With kind regards,
>
> Andrej
>
>
>
> On Sun, Jun 12, 2016 at 12:55 PM, andré hirschowitz <a...@unice.fr> wrote:
> > Hi,
> >
> > I have been thinking to this issue for years (decades?). In France we
> have
> > this research agency INRIA which has been supporting the Coq project for
> > decades, leading to a fairly "good quality software" (in my opinion). Say
> > twenty years ago, the Coq project was targetting the computer science
> > community and was not ready to "attack" the mathematical community. From
> > this side, the picture seems much better nowadays.
> >
> > A possible strategy toward the investment of the mathematical community
> is
> > as follows:
> >
> > ------------------------
> >
> > 1- create a body tying (part of) this community with for instance (part
> of)
> > the Coq project (and/or the Agda project, about which I know little).
> >
> > 2- obtain specific funding from a Research Agency (NSF, CNRS?) for partly
> > formalized PhD fellowships, together with companion funding for the
> > technical support (eg from the Coq team) to the (partial) formalization.
> >
> > 3- obtain good applications, coming from ouside this community.
> >
> > 4- select the winning applications regarding both the interest of the
> naked
> > thesis, and the feasability of the (partial) formalization.
> >
> > 5- help collectively the success of each selected project.
> >
> > 6- Write assessments in particular for the formalization efforts of these
> > newage doctors, so that they win positions whenever they deserve.
> >
> > ---------------------------
> >
> > I leave it here.
> >
> > ah
> >
> >
> >
> >
> > 2016-06-12 10:04 GMT+02:00 Andrej Bauer <andrej...@andrej.com>:
> >>
> >> Apologies for a slightly off topic post, but I think it is relevant to
> >> many people on this list.
> >>
> >> I just looked at some slides by William Stein, the author of Sage (an
> >> open-source alternative to Mathematica) at
> >> http://www.smbc-comics.com/index.php?id=4127
> >>
> >> The conclusion is: it's impossible to make good quality software in
> >> academia because there isn't enough money and because making software
> >> doesn't give one any academic credit.
> >>
> >> I am afraid formalization of math might fall into the same category,
> >> unless we somehow elevate it to a "true science" level. A great deal
> >> has been done in this respect recently by projects lead by Gonthier,
> >> Hales and Voevodsky, but is it enough? Are we even making a dent?
> >>
> >> At my department, for instance, the folk knowledge propagated from one
> >> generation to another is that "someone" formalized "Landau's book" (I
> >> suppose it was the Automath formlaization of Landau's Grundlagen der
> >> Analysis by Jutting) which proves that "it can be done" but is
> >> otherwise an intellectually barren exercise without academic value. I
> >> still remember one of the professors saying this to the whole class
> >> when was an undergraduate.
> >>
> >> With kind regards,
> >>
> >> Andrej
> >>
> >> --
> >> You received this message because you are subscribed to the Google
> Groups
> >> "Homotopy Type Theory" group.
> >> To unsubscribe from this group and stop receiving emails from it, send
> an
> >> email to HomotopyTypeThe...@googlegroups.com.
> >> For more options, visit https://groups.google.com/d/optout.
> >
> >
>
> --
> You received this message because you are subscribed to the Google Groups
> "Homotopy Type Theory" group.
> To unsubscribe from this group and stop receiving emails from it, send an
> email to HomotopyTypeThe...@googlegroups.com.
> For more options, visit https://groups.google.com/d/optout.
>

[-- Attachment #2: Type: text/html, Size: 8362 bytes --]