I suggested that you be invited to join the InChI committee and I hope you join us.

There is no real evidence (afaik) to suggest that CAS tried to get the article pulled. Even if they did, Outsell would not have to oblige and responsibility would still not lie with CAS.

CAS are not the CIA or the KGB (again, afaik).

Equally, so what if the article was ‘unbalanced’? Is it bad practise to publish an article that is positive about InChI? This would surely be useful anyway.

And someone could always post another pointing out its negative aspects, again it would be fine for such an article not to be balanced (whatever that is) for the same reason.

“I’ve seen you support InChI on your blog previously and then of late have entered into the discussions regarding FlexMol and CML…it’s interesting to watch the discussions between yourself and Peter Murray-Rust in this domain and interested parties should watch.”

I neither support or oppose InChI; this shouldn’t be that kind of discussion. Like any technology, InChI has it uses and limitations, and I’ve discussed both for over three years now.

Pointing out InChI’s limitations doesn’t make me an opponent any more than explaining its utility makes me a supporter.

In my reading, the Outsell piece implied that CAS were behind the times in not already generating and storing InChIs in its databases. But there are other plausible explanations that the Outsell piece (and the other discussions I had seen) didn’t even consider, some of which I outlined in my comments above and here:

http://depth-first.com/articles/2009/04/10/why-chemical-abstracts-service-might-not-want-to-use-inchi

It may be tempting to attribute what CAS and Outsell are doing or not doing to some conspiracy theory. But this is ultimately not productive and only obscures the most important question: for which applications is InChI well-suited and for which is it not?

IMO, that’s a discussion that needs to take place – before CAS (or anyone else) rush out to InChI-fy their cheminformatics systems.

(BTW, I’m still getting an error to the effect of ‘page not found’ after posting comments to your blog).

“Just for the record…

CAS manually covers all of chemistry and everything indexed is given a CAS Registry Number.

InChI’s, generated locally (i.e., not in any central location) by individuals and/or individual organizations who care to generate them, can only be generated by a computer algorithm and cover only defined chemical structures (and currently does not cover all classes of chemicals).
“Defined chemical structures” are, by definition, those chemicals whose structure one can draw. This leaves out many, many things covered by CAS, including, as Bob Buntrock points out below, gasoline. Thus InChI’s and the corresponding InChIKey’s are not a replacement or alternative.

Steve”

There are a myriad of technical arguments against InChIs. You have listed some of them. There are a lot more including support of polymers, organometallics, host-guest complexes, excited states, Markush structures, and so on. I’ve seen you support InChI on your blog previously and then of late have entered into the discussions regarding FlexMol and CML…it’s interesting to watch the discussions between yourself and Peter Murray-Rust in this domain and interested parties should watch.

InChI is not perfect. It is the BEST available vendor-neutral international standard that is presently being used by publishers, online databases and cheminformatics vendors to connect and integrate information. Despite its warts it has high value. The development is not over and the InChI developers and subcommittee acknowledge that. The InChI trust are working hard to set up a system that will sustain development costs and effort. I encourage you to participate and bring your ideas and challenges to the committee and sit in!

The acceptance of InChI might be political or technical or a combination of both. We don’t know as there is no commentary from CAS on either aspect. Are you aware of ANY commentary suggesting that it is political or technical. And if so what are your judgments of the other publishers and database groups who are supporting it even with its imperfections?

It IS possible that the Outsell article was one-sided. That said I don’t read it that way and everyone else who has discussed it with me by phone or email seems to agree. Do you think it was one-sided?

- no specification necessary for creating independent implementations;
- single implementation with an open source license that has come under legitimate criticism;
- fails to generate unique identifier for many common molecules (e.g., ferrocene, (R)- and (S)- BINAP);
- InChI itself is problematically long (breaks HTML layouts, for example) for even medium-sized molecules;
- InChIKey has no mechanism for backward-compatibility with newer versions that may fix bugs or add features to the existing implementation;
- the “final” version has only very recently been available.

These may have played a role in the article’s retraction as well. Isn’t it possible that Outsell simply recognized its piece was somewhat one-sided, decided that it didn’t reflect well on on organization in the business of selling thorough research, and pulled it for the reasons they cited?