Comments on: Research tastes better when served with source 100% agree 100% agree

]]>
By: David Roberts/2011/01/24/research-tastes-better-when-served-with-source/#comment-4644 David Roberts Fri, 20 May 2011 09:19:48 +0000 Adding an addendum seems like a good idea - I'll do so when I'm certain that it won't trigger another retweet :) Adding an addendum seems like a good idea – I’ll do so when I’m certain that it won’t trigger another retweet :)

]]>
By: snake5/2011/01/24/research-tastes-better-when-served-with-source/#comment-4636 snake5 Fri, 20 May 2011 05:40:18 +0000 Yeah, I'm very sorry about that. Mike tweeted the link earlier and I assumed it was triggered by a new post, as is usually the case. My bad! :) Feel free to delete my garbage comments and add an addendum instead, or something. Yeah, I’m very sorry about that. Mike tweeted the link earlier and I assumed it was triggered by a new post, as is usually the case. My bad! :)

Feel free to delete my garbage comments and add an addendum instead, or something.

]]>
By: Jonathan Adamczewski/2011/01/24/research-tastes-better-when-served-with-source/#comment-4634 Jonathan Adamczewski Fri, 20 May 2011 05:17:07 +0000 More than just providing research code, people should write clean and well documented reference implementations if they want to get an idea out there. Properly written idiomatic code is an unambiguous and complete representation of the idea it implements (well, complete if you have the entire code) to whoever knows the language. The kind of code that goes into a research implementation usually builds up clutter as ideas are tested and thrown away and it's not written with longevity or clarity in mind. A clean reference implementation would preserve the idea in concrete form much better than standard research code. Another gripe of mine with research papers is the orphaned formula; the formula with very little context, such as how it was derived or what the terms used refer to. This is usually where code is the most helpful, but the paper itself should provide proper context for each formula listed. More than just providing research code, people should write clean and well documented reference implementations if they want to get an idea out there. Properly written idiomatic code is an unambiguous and complete representation of the idea it implements (well, complete if you have the entire code) to whoever knows the language.

The kind of code that goes into a research implementation usually builds up clutter as ideas are tested and thrown away and it’s not written with longevity or clarity in mind. A clean reference implementation would preserve the idea in concrete form much better than standard research code.

Another gripe of mine with research papers is the orphaned formula; the formula with very little context, such as how it was derived or what the terms used refer to. This is usually where code is the most helpful, but the paper itself should provide proper context for each formula listed.

]]>
By: Jonathan Adamczewski/2011/01/24/research-tastes-better-when-served-with-source/#comment-4632 Jonathan Adamczewski Fri, 20 May 2011 05:16:14 +0000 Oh crap, this is an old post. In that case, apologies for the mild snarkiness, plus also misspelling your name! Regardless, the link is well worth following. Oh crap, this is an old post. In that case, apologies for the mild snarkiness, plus also misspelling your name! Regardless, the link is well worth following.

]]>
By: anon/2011/01/24/research-tastes-better-when-served-with-source/#comment-4628 anon Fri, 20 May 2011 05:03:11 +0000

]]>
By: Pavel Shevaev/2011/01/24/research-tastes-better-when-served-with-source/#comment-581 Pavel Shevaev Wed, 26 Jan 2011 10:46:47 +0000 Damn.. Got me! ^^ Damn.. Got me! ^^

]]>
By: James Podesta/2011/01/24/research-tastes-better-when-served-with-source/#comment-579 James Podesta Tue, 25 Jan 2011 12:16:01 +0000 ./altdevblogaday.org/2011/01/24/research-tastes-better-when-served-with-source/#comment-579 you guys just want a playable demo attached to each paper so you don’t need to read the research paper... admit it J<o:p></o:p> From: Posterous [mailto:comment-kugbaalBfHpoIse=posterous.com@sendgrid.me] On Behalf Of Posterous Sent: Tuesday, 25 January 2011 9:52 PM To: podesta1971@gmail.com Subject: [altdevblogaday.com] Comment on "Research tastes better when served with source"<o:p></o:p> you guys just want a playable demo attached to each paper so you don’t need to read the research paper… admit it J<o:p></o:p> From: Posterous [mailto:comment-kugbaalBfHpoIse=posterous.com@sendgrid.me] On Behalf Of Posterous Sent: Tuesday, 25 January 2011 9:52 PM To: podesta1971@gmail.com Subject: [altdevblogaday.com] Comment on "Research tastes better when served with source"<o:p></o:p>

]]>
By: Jonathan Adamczewski/2011/01/24/research-tastes-better-when-served-with-source/#comment-578 Jonathan Adamczewski Tue, 25 Jan 2011 11:51:41 +0000 I agree with you all. All lecture and papers I read could really be separated into "nice ideas but not practicable", and "proven ideas with suggested implementations". I think what Collin mentions is a good point: People do not want to show code, because then the code would be complaint about and not the presented discussed. I think this is a valid suggestion why people do not really provide that much source code (and of course because of the tight limits in pages or words you may write for many publications). I agree with you all. All lecture and papers I read could really be separated into “nice ideas but not practicable”, and “proven ideas with suggested implementations”. I think what Collin mentions is a good point: People do not want to show code, because then the code would be complaint about and not the presented discussed. I think this is a valid suggestion why people do not really provide that much source code (and of course because of the tight limits in pages or words you may write for many publications).

]]>
By: Daniel Collin/2011/01/24/research-tastes-better-when-served-with-source/#comment-576 Daniel Collin Tue, 25 Jan 2011 08:26:11 +0000 I can only say this: YES! I wholeheartly agree. It doesn't even have to compile. Just a small snippet och the relevant parts and a small description of variables declared outside the scope of the snippet. I can only say this: YES! I wholeheartly agree. It doesn’t even have to compile. Just a small snippet och the relevant parts and a small description of variables declared outside the scope of the snippet.

]]>