22
$\begingroup$

I didn't have much time to participate this fall, and I'm excited to see the synthesis golf questions.

Idea:

I'd like to suggest possibilities for “molecular design golf” (name TBD). These are questions that point to the diversity and design aspects of chemistry. Some might include or require computational investigations, but I’d hope that’s not always required for participation.

The key is to propose a question that must be solved with a molecular structure design.

The catch is to ensure the question isn't too broad (e.g., What gas has the highest refractive index?)

So my suggestion is that the question must require some sort of limit, in molecular size or complexity.

Examples:

There have been a few examples that I think would qualify, e.g.:

Meta Question: Is it worth pursuing "molecular design golf" separate from "synthesis golf?" If so, please suggest some possible questions, even if they're not completely well-formed.

$\endgroup$
  • 3
    $\begingroup$ I would welcome that very much, and like in the synthesis golf I am eager to put bounties behind it; this time I might even be able to contribute a challenge. $\endgroup$ – Martin - マーチン Dec 30 '17 at 16:14
  • $\begingroup$ I'd be up for this! The $\ce{C20H18}$ question has gotten a lot of interest (though I actually don't think its a great question as theres no real skill involved in guessing a random non-existent molecule), but if you can come up with something I say just try it and see :) $\endgroup$ – NotEvans. Dec 30 '17 at 16:26
  • 1
    $\begingroup$ I just remembered this from watoc: unusual chemical bonding challenge @ CB2017. It would be a shame to have to wait for something like this for another three years. $\endgroup$ – Martin - マーチン Dec 30 '17 at 16:31
  • 3
    $\begingroup$ @Martin-マーチン - I'm sure I can ask Henry Rzepa for some question suggestions. ;-) $\endgroup$ – Geoff Hutchison Dec 31 '17 at 0:36
  • $\begingroup$ @GeoffHutchison If he didn't already have such a well established blog, I wonder if we could entice him here.... $\endgroup$ – hBy2Py Dec 31 '17 at 4:10
  • $\begingroup$ @hBy2Py try inviting people via CCL, I'm sure there'll be another great discussion. $\endgroup$ – Martin - マーチン Dec 31 '17 at 7:29
  • 1
    $\begingroup$ Interesting proposal. Synthesis golf isn't quite my cup of tea, perhaps mainly because I lack the skill to participate meaningfully (don't get me wrong, I think it's a great initiative, I can see the appeal, and I am blown away with some ingenious reactivity that gets proposed to shorten steps or make use of cheap reagents). I think there is some potential for design challenges to really dig deep into the fundamentals of chemistry, like the polarity question did. Unfortunately, contributing may be quite hard, and may quickly stumble on research-level questions, but it's worth a go! $\endgroup$ – Nicolau Saker Neto Dec 31 '17 at 9:28
  • 2
    $\begingroup$ More generally, this idea reminds me a bit of the concept behind Programming Puzzles & Code Golf. My sense is they developed a lot of their most common rules/limitations as they went along, and saw where the fun-killing shortcuts were. Likely the evolution would be similar for this. Worth a shot, I say! I might even be able to play!! $\endgroup$ – hBy2Py Dec 31 '17 at 13:26
  • 1
    $\begingroup$ I would like to see more of these golf type challenges, but I agree with NotEvans that it should be restricted to existent molecules. I am suggesting questions of the type: finding a molecule with the biggest difference between most upfield and most downfield proton or carbon, or finding a molecule with the biggest difference in partial charges of two of its atoms. $\endgroup$ – RBW Dec 31 '17 at 22:04
  • 2
    $\begingroup$ This is hopefully a lot more doable for those of us who know nothing about synthetic chemistry. $\endgroup$ – pentavalentcarbon Jan 1 '18 at 0:58
  • 1
    $\begingroup$ Concur with @Marko, the objective functions need to be concrete computable quantities, accepting the inevitable tension that computed values do not always match well with experimental ones. A well-posed challenge may well need to specify things like computational method and parameters, possibly the software package to use (presumably free/open-source), and some challenges at least should ideally be constrained to relatively small molecules so that those without significant computational resources can still participate. $\endgroup$ – hBy2Py Jan 1 '18 at 2:58
  • 1
    $\begingroup$ Yes, please! Although, I think the rules need to be formulated with some tact. Like Nicolau already mentioned, I feel it will be very easy to stumble upon open-ended, research-level questions. Although, this doesn't mean we should restrict ourselves to existed molecules. Just my 2 cents. $\endgroup$ – getafix Jan 1 '18 at 6:58
  • $\begingroup$ Great idea, I am an enthusiastic onlooker for code golf (I lack the skill to participate, so the sidelines it is), and am genuinely impressed by the abilities of those who golf. Bring it on. $\endgroup$ – Todd Minehardt Jan 1 '18 at 17:29
7
$\begingroup$

I figure I'll write something since I am enjoying some precious time away from the bench and can actually afford to do so...

As I mention in my comment, this format has great potential to uncover some fundamental aspects of chemistry. That said, I think the success of this format will be rather heavily dependent on the breadth we're willing to accept. In particular, I would like to make a case for the inclusion of questions regarding extremes in chemistry.

Now, I'll admit a large personal bias right from the outset. I love talking and reading about "what is the most/least ____ possible in chemistry?" because it helps establish the boundaries of our Universe and the limits of our knowledge. It's awesome to see how crazy a system can get because someone just really really wanted it to display as much of a certain behaviour as is physically allowed.

I do not think that the fact that extremes are eventually superseded is a good argument against their inclusion. For one, that's what science is all about. There is never a "right" answer, only better and better approximations to reality. No matter what, a significant fraction of answers currently on the site will be superseded or made obsolete in the next few decades (reaction mechanisms, chemical/physical standards, experimental setups, industrial processes, etc); all knowledge has a half-life.

Even when something is superseded, that in no way means the prior information is now useless, especially if we emphasise the journey rather than the destination. Therefore, instead of giving answers like:

  • "this molecule is the highest, see ref here"

, we should gravitate towards something like:

  • "as of 2018, this molecule is the highest known. It makes extensive use of properties X, Y and Z. Property X seems to be at its limit, but it may be possible to exploit Y and Z further. An intuitive argument suggests there must be an upper limit, and detailed theoretical analysis confirms that there can be no example which is twice more than the current experimental limit"

Also, from a purely practical standpoint, the allowance of "extreme" questions immediately provides a huge pool of topics to explore, along with the chemistry within them.

I know there's already been some discussion on this and I wasn't around for it, so I apologise if my arguments have been raised and refuted previously. I am happy to take criticism.

$\endgroup$
  • $\begingroup$ Thanks for the feedback. I agree about accepting "extreme" questions. That's certainly my first proposal - long bonds. I think it'll be a community discussion for a while. $\endgroup$ – Geoff Hutchison Jan 2 '18 at 3:10
4
$\begingroup$

This answer ~reproduces my comments from below the OP, and adds a few more.

This idea reminds me a bit of the concept behind Programming Puzzles & Code Golf. My sense is they developed a lot of their most common rules/limitations as they went along, and saw where the fun-killing shortcuts were. Likely the evolution would be similar for this. Worth a shot, I say! I might even be able to play!!

I concur with Marko, the objective functions need to be concrete computable, calculable, or researchable (e.g., in reference databases) chemical quantities; for computational challenges this would entail accepting the inevitable tension that computed values do not always match well with experimental ones. A well-posed computational challenge may well need to specify things like computational method and parameters, possibly the software package to use (presumably free/open-source), and some challenges at least should ideally be constrained to relatively small molecules so that those without significant computational resources can still participate. Database lookup challenges would ideally constrain answers to come from one or more open data sources. Any answers not conforming to the challenge criteria would then be non-competing.

One other thing to consider implementing as a part of this would be something like the Sandbox for Proposed Challenges implemented at PPCG, where draft challenges are posted for comment by interested members of the community. These sorts of discussions could start here, but there'd be value at least in creating a separate meta post for it, so that discussions on the overarching idea itself would be held separately from the individual discussions on challenge ideas. (Something similar to this is developing with Synthesis Golf, with an overview post living separately from the 'synthesis golf target ideas' post.

$\endgroup$
  • 2
    $\begingroup$ I don't think this should be restricted to "computable" quantities. Questions like finding long C-C single bonds can be approached via crystallographic databases, etc. (People have suggested other cases that could be interesting spectrographic database questions too, e.g., NMR shifts) $\endgroup$ – Geoff Hutchison Jan 1 '18 at 3:39
  • $\begingroup$ And yes, I agree that there would need to be an overview post and a target ideas post as well. $\endgroup$ – Geoff Hutchison Jan 1 '18 at 3:39
  • $\begingroup$ @GeoffHutchison <nod>, point taken. Thoughts on revisions? $\endgroup$ – hBy2Py Jan 1 '18 at 3:53
  • $\begingroup$ @GeoffHutchison Oh, sorry --- I mean, what are your thoughts on the edits I made to my post? :-) Happy New Year! $\endgroup$ – hBy2Py Jan 1 '18 at 4:11
  • $\begingroup$ They look good. Seems like a good start for an overview post. $\endgroup$ – Geoff Hutchison Jan 1 '18 at 4:14
  • $\begingroup$ @GeoffHutchison Thinking more about database searching, from reading Prof. Rzepa's blog, those searches can require a specialist's touch and lots of insight to interpret properly. (Not that computational results are any different, natch.) But, 'this value is scientifically valid' and 'this value exists in a database'/'this value is what the computer spit out' are two very different things, with the former being a much more challenging thing. Which did you have in mind for the mol-fab golf? $\endgroup$ – hBy2Py Jan 1 '18 at 13:24
2
$\begingroup$

I like the idea, and have just bountied the first instalment.

However, I just want to write some of my thoughts. It may or may not be useful, but it is just my 2p.

One has to be clear about the aims of such an initiative. I think that it is certainly fun for people who know how to look for answers. Could it, perhaps, attract new people who may stay and become regular users...? I'm reminded, of course, of which started about half a year ago. Despite objections that it would be too broad I remain a strong supporter of it, because I generally want to raise the quality of Q&A and I think that that would be able to contribute to that larger aim. My view on is similar.

However, as with , I worry that the people who can answer this are few to begin with. syngolf started out fairly strong, but two of the last three challenges have only had one answer. That could be due to my admittedly poor choice of molecule, but also I think that some people who have taken part previously find themselves quite busy. Personally, I do not have either the time or energy to participate fully, especially when my "spare time" on SE consists of moderation work (not actually all that much) and editing work (literally endless). So now the number of people who (1) know how to answer; and (2) are free enough to answer is low. I also perceive a general lack of interest on its meta posts, which is probably because of the same reasons.

So, if this is (partly) about raising quality of Q&A, I wonder if we are doing it the correct way round. Is the time ripe for this kind of initiative - is the userbase sufficiently large? Or is there another way that the quality should be addressed? Obviously, if it is fun, then by all means - there's no harm, apart from the opportunity cost of time.

I remain in support of the idea; this is really just me sharing some insecurities I have, towards syngolf and similar projects. I would be very happy to see the inaugural prove my concerns unfounded.

$\endgroup$
  • $\begingroup$ Incidentally, I've also decided that I won't post a syngolf this month. Let's see how mol design gets along. I guess there are more theoreticians than there are organic chemists here. $\endgroup$ – orthocresol Jan 14 '18 at 3:26
  • 1
    $\begingroup$ I think in both cases, the questions need to be careful in not being too complex. I thought there would be more answers already. My goal is to pick topics that someone could go about an hour or so... this seems a level of complexity in the code golf questions too. $\endgroup$ – Geoff Hutchison Jan 14 '18 at 3:48

You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .