Talk:Info-gap decision theory

Page contents not supported in other languages.
From Wikipedia, the free encyclopedia
WikiProject iconStatistics C‑class Low‑importance
WikiProject icon

This article is within the scope of WikiProject Statistics, a collaborative effort to improve the coverage of statistics on Wikipedia. If you would like to participate, please visit the project page or join the discussion.

CThis article has been rated as C-class on Wikipedia's content assessment scale.
 Low This article has been rated as Low-importance on the importance scale.

ñ and ó instead of – and —[edit]

Can someone please replace ñ and ó with and ? They have been exchanged in this edit: along with some other signs (ö instead of š in Babuška, mutilation of certain quote marks and apostrophes). Unfortunately I seem unable to persuade my browser to do search and replace for me. -- (talk) 13:29, 10 September 2008 (UTC)Reply[reply]

Mojibake fixed; thanks for highlighting the error!
(An alternative is to revert the edit and try again.)
—Nils von Barth (nbarth) (talk) 18:58, 18 March 2009 (UTC)Reply[reply]

Error in Illustrative example[edit]

Figures 2, 5, 6 and 9 could be improved; they are slightly misleading with respect to axis crossing. According to the text ("the right-hand side of the graph corresponds to allocating all resources to the blue team"), the label for the abcissa could be called "Allocation to blue team" and has the range [0, 1]. The graph should not go through the origin, but instead cross the y-axis at the point cooresponding to the return from Figure 1 for allocating all our resources to the red team. Similarly the y-axis on the right side of the graph should correspond to the return from allocating all our resources to the blue team. The general shape of the graph is correct in that the return will increase as more resources are allocated to the blue team until a maximum is reached but thereafter the return will decrease.Dugite (talk) 06:45, 19 November 2008 (UTC) (talk) 06:58, 15 May 2010 (UTC)Reply[reply]

Split & summarize criticism[edit]

I’ve split off the main body of the criticism into a separate article criticism of info-gap decision theory, and focused the criticism on what appear to be the 2 points of contention:

local/global model
Under severe uncertainty, one may argue (as Sneidovich does) that one should not start with a point estimate (local), which is assumed to be very flawed, but rather use the entire space of possible outcomes (global).
This is a substantive point – is info-gap fit for purpose, or misguided?
is it maximax?
Ben-Haim argues that info-gap is not maximax, while Sneidovich argues that it is.
This is an academic point – what should one call it?

The info-gap article, as I understand it, primarily focuses on "what the model is" and "how people use it/advocate its use", with examples; criticism of its use and the nomenclature merit a place, which seems best fulfilled by:

  • fronting some, briefly (in the lede),
  • having a short section elaborating (as above), and
  • having a more extended discussion in a separate article, so as to not obscure the exposition of the main article.

Does people find that the rearrangement and focused criticism seem to improve matters?

—Nils von Barth (nbarth) (talk) 18:49, 18 March 2009 (UTC)Reply[reply]

Oops – I didn’t realize that POV forks violated WP:NPOV; I’ve merged back the removed content, currently in a section called “Alternative presentation”. The article would flow better if the presentation were integrated with the rest of the article, perhaps as another example.
—Nils von Barth (nbarth) (talk) 00:05, 24 March 2009 (UTC)Reply[reply]
This article is still very long - at 133KB, well beyond the maximum recommended size for articles. It would be good if part of it were split out into another article, whether that were the 'Criticism' section or a different one. Robofish (talk) 23:00, 21 February 2010 (UTC)Reply[reply]

There is a lot of web space here devoted to a fringe and unimportant theory. —Preceding unsigned comment added by (talk) 19:07, 18 August 2010 (UTC)Reply[reply]

I completely concur with you. It is a pity that WIKIPEDIA provides a public platform for such theories. In any case, my campaign to contain the spread of this theory, and other unsupported theories in Australia is still on. You'll be surprised to find out how popular this theory is in certain academic circles in Australia! (and other countries such as USA, Canada, Germany, Finland, Japan, Sweden, Norway, Israel, France, Iran, Netherlands) See Sniedo (talk) 02:01, 6 September 2010 (UTC)Reply[reply]

Treatment of severe uncertainty[edit]

The following argument

The logic underlying the above illustration is that the (unknown) true revenue is somewhere in the immediate neighborhood of the (known) estimate of the revenue. For if this is not the case, what is the point of conducting the analysis exclusively in this neighborhood?

is recursive: "Our approach is correct, because, if it weren't, why would we do it this way?" If the purpose is to point out the fallibility of the example approach to severe uncertainty, then it's poorly done. The next two paragraphs aren't great, either.

Can anybody provide a better explanation of how info-gap treats severe uncertainty, hopefully with a brief explanation of what info-gap considers to be "severe" uncertainty? (talk) 07:42, 8 February 2011 (UTC)Reply[reply]

According to info-gap decision theory, severe uncertainty is an uncertainty that is characterized by the following three properties:

  • A vast uncertainty space (e.g. unbounded).
  • A poor point estimate of the parameter of interest (can be substantially wrong).
  • A non-probabilistic, likelihood-free quantification of uncertainty.

Info-gap's robustness model is a re-invention of the Radius of Stability model. That is, according to info-gap decision theory, the robustness of a decision is the smallest perturbation in the point estimate that can destabilize the system. Decisions are ranked according to their robustness, hence the best (optimal) decision is one whose robustness against perturbations in the point estimate is the largest. Sniedo (talk) 07:13, 17 July 2011 (UTC)Reply[reply]

Having issues with instructional language and first/second person pronouns.[edit]

WP:WE - Do not use "we", "let's", or "you". Use "one" or "them" instead. First/second person pronouns and instructional language should only be quotated and attributed to cited reliable sources.

WP:NOTHOWTO - Do not use "should", "must", or "need to". Wikipedia states how or what can, but shouldn't state how to or what should. --Turkeybutt (talk) 11:28, 5 September 2016 (UTC)Reply[reply]

The "we's" and the "shoulds" are in the sources. - DVdm (talk) 11:38, 5 September 2016 (UTC)Reply[reply]
But there are lots of them in this article that are not used in quotation. We have to write it so the article isn't mistaken for an instructional guide or a script for a Ted-ED video. We have to write it so that it puts those instructional and first-second person words into quotations and then the article attributes the quotes to cited reliable sources. --Turkeybutt (talk) 11:45, 5 September 2016 (UTC)Reply[reply]
It does not matter if there are lots. If they are sourced, then Wikipedia can (and probably must) have lots. - DVdm (talk) 11:47, 5 September 2016 (UTC)Reply[reply]
It's just that I don't want instructional facts or factoids about the reader to be on articles just because of cited sources. If the sources contain instructional material or use first-second pronouns, that is no excuse to just plop that stuff on articles like encyclopedic facts. There should be quotation marks in between whatever is being copied from sources. --Turkeybutt (talk) 00:15, 6 September 2016 (UTC)Reply[reply]

Links to Technion pages are outdated[edit]

Current links to Technion sites are not working anymore. This is specifically important for the references, which I don't know how to fix. Some links that are more relevant at this time: — Preceding unsigned comment added by (talk) 23:00, 28 March 2018 (UTC)Reply[reply]