Comments on comments and agents and eyeballs

One of the difficult features of blogs is how to manage comments. On this blog these are relatively infrequent, wile on – say –ChemBark, TotallySynthetic.com or The Chem Blog some articles generate over 100 replies. I got into the habit of responding to most posts, but have sometimes dropped out of the habit, especially when busy.
In the highly commented blogs, I suspect, some of the regulars go back frequently to check how the discussion is going but here, I think, most people visit once. Because of that comments do not always get prominence. So I have also got into the habit of extracting comments and commenting on these.
(BTW in some cases I would much prefer a forum rather than a blog as it allows you to review posts on a given topic more easily. Slashdot uses this type of approach, with added moderation. I am surprised that it isn’t easier to created structured history from a blog – yes, you don’t want it for pictures of kitty (or perhaps you do) – but for blogs with advocacy it’s really valuable. It saves cutting and pasting links and other horrors.)
When I leave a comment unanswered it’s probably simply that I haven’t time. Sometimes also I don’t want to get distracted into a discussion that, though interesting, isn’t mainstream. So here are two recent ones…

1. Steven Bachrach Says:
   October 1st, 2007 at 2:10 pm ePeter,Not to splash cold water on this, but one should keep in mind that the cyclohexinol story blew up because of Rychnovsky’s paper in a closed access journal (Org. Lett.). Ruchnovsky chose not to post this in a blog or even in an OA journal. The blogosphere only then picked up on it, and did greatly promote this subject to a wider audience. I think we are still a long way from web 2.0 being the publication access of choice for chemists for original research. For the time being, web 2.0 seems to be more suited as a more personal, chatty style of communication – sort of a hipper, more broadly authored, and more current version of Chemical and Engineering News (or Chemistry in Britain).Steve
2. ChemSpiderMan Says:
   October 1st, 2007 at 3:55 pm ePeter, I’ve given many examples of the issue of Data Quality on the blog. Some links are:http://www.chemspider.com/blog/?p=64
   http://www.chemspider.com/blog/?p=164
   http://www.chemspider.com/blog/?p=168
   http://www.chemspider.com/blog/?p=137Also, today at the PubChem “Advisory Group” meeting I will be presenting on this issue to the attendees. I will put the presentation online later.
   You should be aware that issues of Quality are showing up already and proliferate problems…for example, while the story about structure drawing quickly is “interesting” the problem is that the structures shown there have errors (at http://liquidcarbon.livejournal.com/13138.html)
   It’s good to know that you appreciate our efforts around Data Quality. I’ve never seen a response to my data quality comments on your blog and I assumed that you would have been very interested in the work I’ve done on Wikipedia and taxol validation this past week. FYI, the link back to PubChem and the systematic name have now been edited on the DrugBox and the Wikipedia record is now correct.

(1). I think I kinda knew this – I am not an organic chemist so I came on the story later. Would it have been discovered by the blogosphere anyway? My take is thusfold:

• if it’s interesting, yes. AFAICS hexacyclinol is not very interesting except in the synthetic organic chemical olympics and so wouldn’t get unearthed except by them. The best I have on the blogosphere is:

• Anonymous (on Tenderbutton’s blog)
  June 1st, 2006 | 12:08 pm
  Has anyone seen Org Lett ASAP today? Take a look at Scott Rychnovsky’s “Predicting NMR Spectra by Computational Methods: Structure Revision of Hexacyclinol” then look at the earlier synthesis paper “Total Synthesis of Hexacyclinol” ACIE 2006 2769-2773 (especially the so-called supplemental data). Things don’t add up. Dylan, can you get to the bottom of this pecular situation?

PMR: so Steve is right – the paper alerted the blogosphere. What would have happened without the public interest? Would it have been reported in C&E news or at least the blog (I think I’m right on that one)? It’s certainly true that quality has a much higher profile because of the discussions.
(2) This is simply because I was too busy. I’m not personally interested in the structures of diazonamide, taxol, etc. and my silence signified “good! thank you”. I will try to add something on a regular basis to comments when possible. I do, however, want to avoid infinite regress of discussion.
…
There are clearly many data “out there” which are “wrong”. Sometimes it is typos – there are a lot in peaklists because it is a particular stupid C19 way or reporting science when there are acceptable (if not brilliant) eSpectra technologies. Some are mislabbeling – right spectrum, wrong compound association. Note that it is the associations that are the problem in all of this, including the versioning. “right” and “wrong” do not apply to molecular formulae or spectra, only the association of names, spectra, molecular structures, etc.
So Le Clair had the wrong association between a structure and a spectrum (and between various samples). This is harder to formalize than it seems. each component is “right”.
One of the real problems is that the association between structure and spectrum is never transmitted – if it were our problems would be much easier. The simple way to do this is using CML – the molecules and spectra are contained in the same container. NMRShiftDB does this, and we have been able to download thousands of CML files, extract the structures and spectra (as peaklists) automatically.
Now what could we possibly want to do that for? If you comment I guarantee to comment – once.
… and did I mention “Open data”?