From: Martin Escardo <m.escardo@cs.bham.ac.uk>
To: Patrik Eklund <peklund@cs.umu.se>, Categories <categories@mta.ca>
Subject: Re: Current Issues in the Philosophy of Practice of Mathematics & Informatics
Date: Thu, 30 Jul 2015 15:46:26 +0100 [thread overview]
Message-ID: <E1ZLtMv-0007rT-7G@mlist.mta.ca> (raw)
In-Reply-To: <E1ZKnoO-0002MU-Si@mlist.mta.ca>
Some claims quoted below need to be rectified, given that this is a
public forum:
On 29/07/15 06:54, Patrik Eklund wrote:
>> Yes, we cannot create the set of all sets, similarly as we shouldn't
> even try out creating the type of all types. Nevertheless, Martin-L??f
> took the liberty of doing that, and was opportunistic enough to publish
> it. Things went wrong but it was not called a paradox.
It is called Girard's Paradox, and the construction resembles
Burali-Forti's Paradox, rather than Russell's paradox.
The idea was to have a type U of all types, including U itself,
written U:U. This may seem naive given Russell's Paradox was known.
However, there is more to U:U than Russell's paradox, because "U:U" is
not a proposition (it is a so-called judgment), and hence it cannot be
true or false, or taken as a hypothesis in a mathematical
statement. In particular, using U:U, you cannot form the type of all
types that don't belong to themselves, because there is no "belong"
relation in type theory, and for instance writing something such as
"not(X:X)" is not even grammatically correct.
To derive a contradiction using U:U (in a type theory extended with
this judgement) is much harder than to derive a contradiction from the
hypothetical existence of a set of all sets (in set theory).
> Constructions were "improved" over decades, but the HoTT community
> still uses universality, so that paradox just appears as the
> emperors new clothes.
The improvement adopted both in MLTT in the 1980's, and in MLTT+HoTT
axioms now, was already available, and is the same as the one Russell
proposed a century ago to avoid his paradox, and adopted in Principia
Mathematica, namely to instead have a hierarchical stratification
U_0 : U_1 : U_2 : U_3 : ... by "size", where U_0 is the type of all
small types, which lives in the type U_1 of large types, which lives
in the type U_2 of even larger types, etc.
The idea is at least 107 years old.
M.
[For admin and other information see: http://www.mta.ca/~cat-dist/ ]
next prev parent reply other threads:[~2015-07-30 14:46 UTC|newest]
Thread overview: 15+ messages / expand[flat|nested] mbox.gz Atom feed top
2015-07-24 9:12 Ralph Matthes
2015-07-25 13:57 ` Graham White
2015-07-26 15:33 ` Patrik Eklund
2015-07-29 1:42 ` Martin Escardo
[not found] ` <55B82F7F.60302@cs.bham.ac.uk>
2015-07-29 5:54 ` Patrik Eklund
2015-07-30 14:46 ` Martin Escardo [this message]
2015-07-31 10:35 ` Thomas Streicher
2015-07-29 13:56 ` Robert Dawson
2015-07-31 5:10 ` Vaughan Pratt
2015-08-04 15:45 ` Patrik Eklund
2015-08-09 2:10 Fred E.J. Linton
[not found] <536THicJV0416S02.1439086221@web02.cms.usa.net>
2015-08-09 9:52 ` Patrik Eklund
2015-08-11 9:12 ` Thomas Streicher
2015-08-11 9:39 ` Steve Vickers
2015-08-11 12:20 ` Robert Dawson
Reply instructions:
You may reply publicly to this message via plain-text email
using any one of the following methods:
* Save the following mbox file, import it into your mail client,
and reply-to-all from there: mbox
Avoid top-posting and favor interleaved quoting:
https://en.wikipedia.org/wiki/Posting_style#Interleaved_style
* Reply using the --to, --cc, and --in-reply-to
switches of git-send-email(1):
git send-email \
--in-reply-to=E1ZLtMv-0007rT-7G@mlist.mta.ca \
--to=m.escardo@cs.bham.ac.uk \
--cc=categories@mta.ca \
--cc=peklund@cs.umu.se \
/path/to/YOUR_REPLY
https://kernel.org/pub/software/scm/git/docs/git-send-email.html
* If your mail client supports setting the In-Reply-To header
via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line
before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).