[Top][All Lists]

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: [H-source-users] better analysis/presentation?

From: Yuchen Pei
Subject: Re: [H-source-users] better analysis/presentation?
Date: Fri, 05 Nov 2021 23:07:50 +1100
User-agent: mu4e 1.4.13; emacs 27.2

Denis 'GNUtoo' Carikli <GNUtoo@cyberdimension.org> writes:

On Tue, 2 Nov 2021 19:23:16 -0400
bill-auger <bill-auger@peers.community> wrote:

On Mon, 01 Nov 2021 21:09:47 +1100 Yuchen wrote:
> Can you clarify with an example what you mean by conflicting > entries?
On Mon, 01 Nov 2021 21:09:47 +1100 Yuchen wrote:
> OK rereading your message I think you mean something like
> > > what happens when another client
> >  submits a contradictory report on the same release of the
> > singular 'guix' distro
multiple versions of distros is a separate issue - the first
is simple - i meant "conflicts" in the plain sense - does a
search result for "foomatic-123" present "Works with free
software? Yes"? , "No"? , or both (in conflicting per-user

alice posts data including a common hardware (eg: i915):

  model = foomatic-123
  works_with_free_software = Yes

bob's computer or distro disagrees:

  model = foomatic-123
  works_with_free_software = No

presumably, a search will result in both, and they will
contradict - furthermore, the burden of analyzing the data-set
is on each user - this suggests some interesting analysis
questions, which could be resolved programatically, if it is not

  how many other users, agree with alice and/or bob?

  is it possible to present search results as a compact table?

    eg: "Foomatic-2000"
"Alice | Works with free software? Yes (for me on trisquel9)" "Bob | Works with free software? No (for me on trisquel7)" "Carol | Works with free software? Yes (for me on parabola)"

  is it is possible to collate/average entries common hardware,
  into a [SIC]"user-confidence-rating"?

    eg: "Works with free software? Yes (41/42 users)"

maybe some otherwise hidden wisdom may be revealed with better
presentation; such as: all 'Yes' were also tested the same one
distro, and zero 'No' were tested on that distro - all 'No' were
from another distro
We could generalize it this way: we take only the best grades (Yes, Grade A) as a summary, but we could warn that it doesn't work in this
way in all cases.

Isn't it better to make the summary the mode / average / median rather than "max"?

For instance that printer is Yes, Grade A, but has a No, we could add a big warning that says that it didn't work for everybody, and encourage
people to check the details.


Here if Ramona wants to know if that printer works, she'll look at the entries and since she runs Trisquel 9, she'll see that it's probably
fine for her.


This would probably also encourage users to fix bogus entries, if someone has an intel WiFi card that somehow works in Trisquel 9, people will probably be pushed to verify that's it's really the case and fix it if it's not. Here we'd also need to have a way to declare entries as 'needing verification' or something like that in very suspicious cases.

This would be useful feature indeed but how do we mark something as suspicious, manually by conventional wisdom (like the example you give but then who would have the privilege to do so and how can we make it explanable), or automatically (but how)?



PGP Key: 47F9 D050 1E11 8879 9040  4941 2126 7E93 EF86 DFD0

Attachment: signature.asc
Description: PGP signature

reply via email to

[Prev in Thread] Current Thread [Next in Thread]