> One concrete suggestion to make the foundations of > coev. algorithms more widely considered (in particular, > better known to IEEE people, if there is such a concept!) might > be a special issue of IEEE TEC on coevolution. Any > volunteers for guest editors?
I am reluctant to take up another service project, as I already have my hands full with FOGA ... and that's over a year away. But *AM* interested in discussing the idea of a special issue, even if I am not willing to edit it. 8^)
Whenever someone mentions the possibility of a special issue, I immediately think: Do I have anything to submit to such an issue? And, of course, I think I do. But my work is (typically) theoretical in nature and I have a relatively oversimplified hash in my mind placing theory in the TCS/ECJ bucket and more applied, empirical work in the IEEE bucket.
Moreover, one of my major concerns is not just raising the scope of CoEC, but in particular raising the scope of the newer perspectives on CoEC made possible by recent theoretical advances. I'm just not sure IEEE TEC is the right venue for that.
Am I being unfair?
Would it be unfair of me to use this IEEE Google Group to posit other possibilities? For example, the new Natural Computating journal, or even ECJ. Better yet: I wonder if there's a chance we could get a special issue in TCSC?
Additionally, a general survey, of the four people that regularly read this: How many would have something they would want to submit to a special issue in the near future?
Hm - I don't know whether I count as one of the four! I expect I am not sufficiently engaged in CoEC research to offer up a quality contribution, though I'd be interested in working with someone on some kind of survey paper, perhaps on games, or on multi-objective. I would hope that others, not necessarily in this group, would have plenty to say.
Are you being unfair? Yes, I think maybe so! It's true that in general, TEC seems to lean towards empirical work, but not exclusively so. And personally, I'd like to see more theory papers published there and a greater awareness of theory shown in all papers. Such a change would have to start somewhere. Why not with this special issue?
phi
On Aug 3, 11:49 pm, "R. Paul Wiegand" <p...@tesseract.org> wrote:
> > One concrete suggestion to make the foundations of > > coev. algorithms more widely considered (in particular, > > better known to IEEE people, if there is such a concept!) might > > be a special issue of IEEE TEC on coevolution. Any > > volunteers for guest editors?
> I am reluctant to take up another service project, as I already have > my hands full with FOGA ... and that's over a year away. But *AM* > interested in discussing the idea of a special issue, even if I am not > willing to edit it. 8^)
> Whenever someone mentions the possibility of a special issue, I > immediately think: Do I have anything to submit to such an issue? > And, of course, I think I do. But my work is (typically) theoretical > in nature and I have a relatively oversimplified hash in my mind > placing theory in the TCS/ECJ bucket and more applied, empirical work > in the IEEE bucket.
> Moreover, one of my major concerns is not just raising the scope of > CoEC, but in particular raising the scope of the newer perspectives on > CoEC made possible by recent theoretical advances. I'm just not sure > IEEE TEC is the right venue for that.
> Am I being unfair?
> Would it be unfair of me to use this IEEE Google Group to posit other > possibilities? For example, the new Natural Computating journal, or > even ECJ. Better yet: I wonder if there's a chance we could get a > special issue in TCSC?
> Additionally, a general survey, of the four people that regularly read > this: How many would have something they would want to submit to a > special issue in the near future?
> I don't know whether I count as one of the four!
I was only joking that there seems to be a small number of people contributing to the discussion in the Google group. (You are involved in the discussions so you would be one of the "four" to which I was referring 8^) ).
> Are you being unfair? Yes, I think maybe so! It's true that in > general, TEC seems to lean towards empirical work, but not exclusively > so. And personally, I'd like to see more theory papers published there > and a greater awareness of theory shown in all papers. Such a change > would have to start somewhere. Why not with this special issue?
Fair enough. I'll give it some thought. Of course, it may in large part depend on who picks up the baton and pushes for a special issue.
Of course, there's an advantage in TCSC, as well ... a wider audience, a pretty strong level of respectability in the CS community in general. I'm not saying TEC is not respectable, just that there may be (perhaps) a broader community in which TCSC is respected. But all this is speculation.
Anthony asked: "Paul, are you involved with organizing the next FOGA? "
(A shameless plug:) Yes. The next FOGA will in be January 2009 in Orlando. Annie Wu, Thomas Jansen, Ivan Garibay, and I are among those putting it together. Web page and such coming soon ...
It hadn't occurred to me to characterize ECJ and TEC as theory and applications journals, respectively. Looking over their contents in a cursory way, perhaps this trend does exist somewhat. But, I agree with Phi that there's plenty to theory published in TEC; let's not forget that the NFL papers are both published there! And, TEC consistently has very high impact factors.
As for a special issue on coevolution, TEC did have a special issue on games 9/6 (2005), which did have a few papers about coevolution; yet other papers used coevolution, but were about other things, like evolution of coooperation. So perhaps that 2005 special issue doesn't make a new coevolution special issue inappropriate. I would be able to contribute something to such a special issue, too. I'm open to co- editing, also, conditioned on things like when we plan to do this, etc..
With respect to TCS, if we want "modern" coevolutionary work to gain wider acceptance in the EC community, it might be that TCS is too remote in topic to help with that. One can argue that it will help promote coevolution to a wider audience, however. I guess we need to decide who we want our target audience to be, and the purpose behind selecting that target audience. If we decide to focus on the EC community, to begin with, then perhaps we want an EC journal (i.e., ECJ or TEC). I'd make a similar argument about Natural Computing -- we need to decide who we want our audience to be and why, and then go after the appropriate journal. Also, consider that if propose a special issue for TCS, for example, who of their regular readership is going to submit? To get papers, we're going to have to advertise in the EC community, probably; the set of contributors may be disjoint from the set of regular readers.
To summarize:
1) other threads and discussions have expressed a frustration that, while a) coevolution is viewed as being important, and b) coevolution events (tracks, tutorials) are consistently well attended, we seem to have a problem A) bringing new people into the field, B) getting new ideas adopted by the community of people using coevolution.
2) to help with this problem, ideas that have been proposed include a) more tutorials, b) competitions, c) special issues, d) special coevolution-focused workshops outside of CEC and gecco, e) killer apps for the new ideas, and so on.
3) if we want coevolution (and modern coevolution) ideas to get more widely adopted, and we want to do a special issue, what community of researchers do we want to begin with? Do we want to start with A) people using coevolution, but with older methods? B) EC people generally, who at least have heard of coevolution, may think it is important, and likely know that we are active in coevolution? C) people more outside of EC, who come to coev tabula rasa and don't have biases towards older methods?
Also, to promote the modern coevolution ideas, it seems like we'll have to be the ones to do that; so, a special issue without (or with few) contributions from us will certainly help promote coevolution, but it might not advance the state of acceptance of the newer ideas. Yet, a special issue with mostly contributions from us might be a bit insular.
If no one else will agree to help Sevan co-edit a special issue, I would do so (with similar conditions as Sevan notes); however, it is spreading myself a bit thin ... so I'll happily yield that honor to someone else who wants to "step up"?
As to point number three, the ideal would be to attract some distribution encompassing all three categories Sevan mentions, but I don't think that will be possible. The crossover appeal of each of these journals is probably quite limited ... though I speculate Nat. Comp. might have a fairly interesting demographic in terms of interest.
What if we ask ourselves a slightly different question: Which of the three categories offers the most potential in terms of numbers of eyes looking at the papers and likelihood to draw interest? Without reason or evidence, I guess (B). The more narrow group of coev people is probably too small to expect much, and the broader group probably will not look closely at the papers due to lack of sufficient interest. That tilts things toward ECJ, TEC, or GPEM (sorry, I forgot to list GPEM in my previous message).
Maybe this is a silly choice to debate before approaching various journals. Some journals may not be interested in such a special issue, and others may have time-lines that don't work for us. Perhaps we could produce a short-list, iron out some high-level details (time- frame, specific topic), then approach the journals on our short list for interest? Constraints may remove choices.
As to the inbreeding problem, Sevan is bang-on. This is a real problem, I think. I brought up the issue of having something to submit only to determine whether there would even be sufficient content to put an issue out in the first place. Obviously, it would be good to have as many submissions to such an issue as is possible, from as many different sources as possible. That might inform our topic ... a more focused topic would be useful from an advocacy point of view (i.e., modern coevolutionary foundations), but it would certainly narrow the pool of potential submissions. On the other hand, a broader topic (e.g., "coevolution") might draw a lot more people but might run the risk of diluting a good opportunity.
Would it be over the top to mine the last few conference proceedings for coev researchers, then email them to see if they would be interested in submitting to such a special issue?
Personally I'm torn between two aims. One is to broadcast what's been done; the other is to advance the limits of what is known. I tend to think that workshops, tutorials, discussion groups, courses, and books are better venues for disseminating ideas than a special issue of a journal would be. Journals are where you put research results. A book which found its way into the hands of new researchers or was used to teach special topics in courses would spread the word more reliably, wouldn't it?
It's true that a result with a high "wow" factor published in a conference or journal can bring a lot of attention to the field. But to the extent that such results are paradigm shifts, we cannot sit around waiting for one to appear or expect that our own work is going to result in one any time soon (it's not easy to predict what's going to capture people's attention, in any case). In the meantime, I think persistence is the only strategy, so we continue to hold and improve tutorials and other such events. We consistently have good attendance at these, and if we're persistent enough the ideas are bound to spread, aren't they? It may be frustrating, but so is sanding wood. I've spent a bunch of time thinking and talking about why the average EC person is stuck in 1998 when it comes to thinking about coevolution, but I still have no idea why. So I keep thinking that we should adjust the tutorials each year to try out new ways of presenting the information until we find the magic formula.
Since special journal issues on coevolution are so few and far between, I personally would like to send some new (read: unpublished) results to one. After all, those of us actively researching coevolution need a venue for publishing new work, don't we? I don't think there's anything wrong with the dialog among active researchers outpacing what the community at large can easily follow. I suppose there's a fine line between insular and communal, but the field will stop going anywhere if the people who actively advance it spend all their time trying to get everyone else to understand what was done five years ago.
I'd happily co-edit, or even guest edit outright, a special issue on coev.
Sorry about beating a dead horse with so many words. I wonder if Jordan would be interested in doing that.
Anthony
On Aug 12, 12:33 pm, se...@eecs.harvard.edu wrote:
> It hadn't occurred to me to characterize ECJ and TEC as theory and > applications journals, respectively. Looking over their contents in a > cursory way, perhaps this trend does exist somewhat. But, I agree with > Phi that there's plenty to theory published in TEC; let's not forget > that the NFL papers are both published there! And, TEC consistently > has very high impact factors.
> As for a special issue on coevolution, TEC did have a special issue on > games 9/6 (2005), which did have a few papers about coevolution; yet > other papers used coevolution, but were about other things, like > evolution of coooperation. So perhaps that 2005 special issue doesn't > make a new coevolution special issue inappropriate. I would be able to > contribute something to such a special issue, too. I'm open to co- > editing, also, conditioned on things like when we plan to do this, > etc..
> With respect to TCS, if we want "modern" coevolutionary work to gain > wider acceptance in the EC community, it might be that TCS is too > remote in topic to help with that. One can argue that it will help > promote coevolution to a wider audience, however. I guess we need to > decide who we want our target audience to be, and the purpose behind > selecting that target audience. If we decide to focus on the EC > community, to begin with, then perhaps we want an EC journal (i.e., > ECJ or TEC). I'd make a similar argument about Natural Computing -- we > need to decide who we want our audience to be and why, and then go > after the appropriate journal. Also, consider that if propose a > special issue for TCS, for example, who of their regular readership is > going to submit? To get papers, we're going to have to advertise in > the EC community, probably; the set of contributors may be disjoint > from the set of regular readers.
> To summarize:
> 1) other threads and discussions have expressed a frustration that, > while a) coevolution is viewed as being important, and b) coevolution > events (tracks, tutorials) are consistently well attended, we seem to > have a problem A) bringing new people into the field, B) getting new > ideas adopted by the community of people using coevolution.
> 2) to help with this problem, ideas that have been proposed include a) > more tutorials, b) competitions, c) special issues, d) special > coevolution-focused workshops outside of CEC and gecco, e) killer apps > for the new ideas, and so on.
> 3) if we want coevolution (and modern coevolution) ideas to get more > widely adopted, and we want to do a special issue, what community of > researchers do we want to begin with? Do we want to start with A) > people using coevolution, but with older methods? B) EC people > generally, who at least have heard of coevolution, may think it is > important, and likely know that we are active in coevolution? C) > people more outside of EC, who come to coev tabula rasa and don't have > biases towards older methods?
> Also, to promote the modern coevolution ideas, it seems like we'll > have to be the ones to do that; so, a special issue without (or with > few) contributions from us will certainly help promote coevolution, > but it might not advance the state of acceptance of the newer ideas. > Yet, a special issue with mostly contributions from us might be a bit > insular.
Dang, one of my sentences jumped to the wrong paragraph. I emailed Jordan asking if he'd be interested in editing a journal issue on coev. T Awhile back I'd emailed Marc Schoenauer about having a special issue of ECJ but he never wrote me back. I assembled a list of potential contributors and started into writing a CFP, so maybe we can use some of that.
On Aug 21, 11:13 am, abucci <abu...@gmail.com> wrote:
> Personally I'm torn between two aims. One is to broadcast what's been > done; the other is to advance the limits of what is known. I tend to > think that workshops, tutorials, discussion groups, courses, and books > are better venues for disseminating ideas than a special issue of a > journal would be. Journals are where you put research results. A > book which found its way into the hands of new researchers or was used > to teach special topics in courses would spread the word more > reliably, wouldn't it?
> It's true that a result with a high "wow" factor published in a > conference or journal can bring a lot of attention to the field. But > to the extent that such results are paradigm shifts, we cannot sit > around waiting for one to appear or expect that our own work is going > to result in one any time soon (it's not easy to predict what's going > to capture people's attention, in any case). In the meantime, I think > persistence is the only strategy, so we continue to hold and improve > tutorials and other such events. We consistently have good attendance > at these, and if we're persistent enough the ideas are bound to > spread, aren't they? It may be frustrating, but so is sanding wood. > I've spent a bunch of time thinking and talking about why the average > EC person is stuck in 1998 when it comes to thinking about > coevolution, but I still have no idea why. So I keep thinking that we > should adjust the tutorials each year to try out new ways of > presenting the information until we find the magic formula.
> Since special journal issues on coevolution are so few and far > between, I personally would like to send some new (read: unpublished) > results to one. After all, those of us actively researching > coevolution need a venue for publishing new work, don't we? I don't > think there's anything wrong with the dialog among active researchers > outpacing what the community at large can easily follow. I suppose > there's a fine line between insular and communal, but the field will > stop going anywhere if the people who actively advance it spend all > their time trying to get everyone else to understand what was done > five years ago.
> I'd happily co-edit, or even guest edit outright, a special issue on > coev.
> Sorry about beating a dead horse with so many words. I wonder if > Jordan would be interested in doing that.
> Anthony
> On Aug 12, 12:33 pm, se...@eecs.harvard.edu wrote:
> > Hi All,
> > It hadn't occurred to me to characterize ECJ and TEC as theory and > > applications journals, respectively. Looking over their contents in a > > cursory way, perhaps this trend does exist somewhat. But, I agree with > > Phi that there's plenty to theory published in TEC; let's not forget > > that the NFL papers are both published there! And, TEC consistently > > has very high impact factors.
> > As for a special issue on coevolution, TEC did have a special issue on > > games 9/6 (2005), which did have a few papers about coevolution; yet > > other papers used coevolution, but were about other things, like > > evolution of coooperation. So perhaps that 2005 special issue doesn't > > make a new coevolution special issue inappropriate. I would be able to > > contribute something to such a special issue, too. I'm open to co- > > editing, also, conditioned on things like when we plan to do this, > > etc..
> > With respect to TCS, if we want "modern" coevolutionary work to gain > > wider acceptance in the EC community, it might be that TCS is too > > remote in topic to help with that. One can argue that it will help > > promote coevolution to a wider audience, however. I guess we need to > > decide who we want our target audience to be, and the purpose behind > > selecting that target audience. If we decide to focus on the EC > > community, to begin with, then perhaps we want an EC journal (i.e., > > ECJ or TEC). I'd make a similar argument about Natural Computing -- we > > need to decide who we want our audience to be and why, and then go > > after the appropriate journal. Also, consider that if propose a > > special issue for TCS, for example, who of their regular readership is > > going to submit? To get papers, we're going to have to advertise in > > the EC community, probably; the set of contributors may be disjoint > > from the set of regular readers.
> > To summarize:
> > 1) other threads and discussions have expressed a frustration that, > > while a) coevolution is viewed as being important, and b) coevolution > > events (tracks, tutorials) are consistently well attended, we seem to > > have a problem A) bringing new people into the field, B) getting new > > ideas adopted by the community of people using coevolution.
> > 2) to help with this problem, ideas that have been proposed include a) > > more tutorials, b) competitions, c) special issues, d) special > > coevolution-focused workshops outside of CEC and gecco, e) killer apps > > for the new ideas, and so on.
> > 3) if we want coevolution (and modern coevolution) ideas to get more > > widely adopted, and we want to do a special issue, what community of > > researchers do we want to begin with? Do we want to start with A) > > people using coevolution, but with older methods? B) EC people > > generally, who at least have heard of coevolution, may think it is > > important, and likely know that we are active in coevolution? C) > > people more outside of EC, who come to coev tabula rasa and don't have > > biases towards older methods?
> > Also, to promote the modern coevolution ideas, it seems like we'll > > have to be the ones to do that; so, a special issue without (or with > > few) contributions from us will certainly help promote coevolution, > > but it might not advance the state of acceptance of the newer ideas. > > Yet, a special issue with mostly contributions from us might be a bit > > insular.
I agree that a book might be a better way to spread the word than a special issue, but at least the special issue might be a way to stimulate participation in coevolution research.
I agree that persistence is necessary to get the word out -- it will take a little time. I think older coevolution heuristics are more commonly used for a few reasons: 1) they've been around longer and so have had more time to be picked up by researchers, 2) they are typically quite simple to implement, 3) several (e.g., Rosin's) heuristics have been demonstrated to be helpful in many circumstances.
On the other side, more recent approaches are beginning to get picked up, as well. Heuristics like LAPCA and the Nash memory have been used by other researchers, for example.
> Personally I'm torn between two aims. One is to broadcast what's been > done; the other is to advance the limits of what is known. I tend to > think that workshops, tutorials, discussion groups, courses, and books > are better venues for disseminating ideas than a special issue of a > journal would be. Journals are where you put research results. A > book which found its way into the hands of new researchers or was used > to teach special topics in courses would spread the word more > reliably, wouldn't it?
> It's true that a result with a high "wow" factor published in a > conference or journal can bring a lot of attention to the field. But > to the extent that such results are paradigm shifts, we cannot sit > around waiting for one to appear or expect that our own work is going > to result in one any time soon (it's not easy to predict what's going > to capture people's attention, in any case). In the meantime, I think > persistence is the only strategy, so we continue to hold and improve > tutorials and other such events. We consistently have good attendance > at these, and if we're persistent enough the ideas are bound to > spread, aren't they? It may be frustrating, but so is sanding wood. > I've spent a bunch of time thinking and talking about why the average > EC person is stuck in 1998 when it comes to thinking about > coevolution, but I still have no idea why. So I keep thinking that we > should adjust the tutorials each year to try out new ways of > presenting the information until we find the magic formula.
> Since special journal issues on coevolution are so few and far > between, I personally would like to send some new (read: unpublished) > results to one. After all, those of us actively researching > coevolution need a venue for publishing new work, don't we? I don't > think there's anything wrong with the dialog among active researchers > outpacing what the community at large can easily follow. I suppose > there's a fine line between insular and communal, but the field will > stop going anywhere if the people who actively advance it spend all > their time trying to get everyone else to understand what was done > five years ago.
> I'd happily co-edit, or even guest edit outright, a special issue on > coev.
> Sorry about beating a dead horse with so many words. I wonder if > Jordan would be interested in doing that.
> Anthony
> On Aug 12, 12:33 pm, se...@eecs.harvard.edu wrote: >> Hi All,
>> It hadn't occurred to me to characterize ECJ and TEC as theory and >> applications journals, respectively. Looking over their contents in a >> cursory way, perhaps this trend does exist somewhat. But, I agree >> with >> Phi that there's plenty to theory published in TEC; let's not forget >> that the NFL papers are both published there! And, TEC consistently >> has very high impact factors.
>> As for a special issue on coevolution, TEC did have a special >> issue on >> games 9/6 (2005), which did have a few papers about coevolution; yet >> other papers used coevolution, but were about other things, like >> evolution of coooperation. So perhaps that 2005 special issue doesn't >> make a new coevolution special issue inappropriate. I would be >> able to >> contribute something to such a special issue, too. I'm open to co- >> editing, also, conditioned on things like when we plan to do this, >> etc..
>> With respect to TCS, if we want "modern" coevolutionary work to gain >> wider acceptance in the EC community, it might be that TCS is too >> remote in topic to help with that. One can argue that it will help >> promote coevolution to a wider audience, however. I guess we need to >> decide who we want our target audience to be, and the purpose behind >> selecting that target audience. If we decide to focus on the EC >> community, to begin with, then perhaps we want an EC journal (i.e., >> ECJ or TEC). I'd make a similar argument about Natural Computing >> -- we >> need to decide who we want our audience to be and why, and then go >> after the appropriate journal. Also, consider that if propose a >> special issue for TCS, for example, who of their regular >> readership is >> going to submit? To get papers, we're going to have to advertise in >> the EC community, probably; the set of contributors may be disjoint >> from the set of regular readers.
>> To summarize:
>> 1) other threads and discussions have expressed a frustration that, >> while a) coevolution is viewed as being important, and b) coevolution >> events (tracks, tutorials) are consistently well attended, we seem to >> have a problem A) bringing new people into the field, B) getting new >> ideas adopted by the community of people using coevolution.
>> 2) to help with this problem, ideas that have been proposed >> include a) >> more tutorials, b) competitions, c) special issues, d) special >> coevolution-focused workshops outside of CEC and gecco, e) killer >> apps >> for the new ideas, and so on.
>> 3) if we want coevolution (and modern coevolution) ideas to get more >> widely adopted, and we want to do a special issue, what community of >> researchers do we want to begin with? Do we want to start with A) >> people using coevolution, but with older methods? B) EC people >> generally, who at least have heard of coevolution, may think it is >> important, and likely know that we are active in coevolution? C) >> people more outside of EC, who come to coev tabula rasa and don't >> have >> biases towards older methods?
>> Also, to promote the modern coevolution ideas, it seems like we'll >> have to be the ones to do that; so, a special issue without (or with >> few) contributions from us will certainly help promote coevolution, >> but it might not advance the state of acceptance of the newer ideas. >> Yet, a special issue with mostly contributions from us might be a bit >> insular.
Jordan said he'd guest edit a special issue, too. So, we have lots of people willing to work on it. All we have to do is decide what's in it. -)
He also pointed out that it doesn't look good if most of the papers in the issue are written by the guest editor(s).
I'm wondering about turning the question around and seeing what other people are interested in writing about coevolution. We all have our preferences about what we'd like to write or read, but what about other people, the ones we're hoping to reach? What would they want to submit?
Anthony
On 8/21/07, Sevan G. Ficici <se...@eecs.harvard.edu> wrote:
> I agree that a book might be a better way to spread the word than a > special issue, but at least the special issue might be a way to > stimulate participation in coevolution research.
> I agree that persistence is necessary to get the word out -- it will > take a little time. I think older coevolution heuristics are more > commonly used for a few reasons: 1) they've been around longer and so > have had more time to be picked up by researchers, 2) they are > typically quite simple to implement, 3) several (e.g., Rosin's) > heuristics have been demonstrated to be helpful in many circumstances.
> On the other side, more recent approaches are beginning to get picked > up, as well. Heuristics like LAPCA and the Nash memory have been used > by other researchers, for example.
> -Sevan
> On Aug 21, 2007, at 11:13 AM, abucci wrote:
> > Personally I'm torn between two aims. One is to broadcast what's been > > done; the other is to advance the limits of what is known. I tend to > > think that workshops, tutorials, discussion groups, courses, and books > > are better venues for disseminating ideas than a special issue of a > > journal would be. Journals are where you put research results. A > > book which found its way into the hands of new researchers or was used > > to teach special topics in courses would spread the word more > > reliably, wouldn't it?
> > It's true that a result with a high "wow" factor published in a > > conference or journal can bring a lot of attention to the field. But > > to the extent that such results are paradigm shifts, we cannot sit > > around waiting for one to appear or expect that our own work is going > > to result in one any time soon (it's not easy to predict what's going > > to capture people's attention, in any case). In the meantime, I think > > persistence is the only strategy, so we continue to hold and improve > > tutorials and other such events. We consistently have good attendance > > at these, and if we're persistent enough the ideas are bound to > > spread, aren't they? It may be frustrating, but so is sanding wood. > > I've spent a bunch of time thinking and talking about why the average > > EC person is stuck in 1998 when it comes to thinking about > > coevolution, but I still have no idea why. So I keep thinking that we > > should adjust the tutorials each year to try out new ways of > > presenting the information until we find the magic formula.
> > Since special journal issues on coevolution are so few and far > > between, I personally would like to send some new (read: unpublished) > > results to one. After all, those of us actively researching > > coevolution need a venue for publishing new work, don't we? I don't > > think there's anything wrong with the dialog among active researchers > > outpacing what the community at large can easily follow. I suppose > > there's a fine line between insular and communal, but the field will > > stop going anywhere if the people who actively advance it spend all > > their time trying to get everyone else to understand what was done > > five years ago.
> > I'd happily co-edit, or even guest edit outright, a special issue on > > coev.
> > Sorry about beating a dead horse with so many words. I wonder if > > Jordan would be interested in doing that.
> > Anthony
> > On Aug 12, 12:33 pm, se...@eecs.harvard.edu wrote: > >> Hi All,
> >> It hadn't occurred to me to characterize ECJ and TEC as theory and > >> applications journals, respectively. Looking over their contents in a > >> cursory way, perhaps this trend does exist somewhat. But, I agree > >> with > >> Phi that there's plenty to theory published in TEC; let's not forget > >> that the NFL papers are both published there! And, TEC consistently > >> has very high impact factors.
> >> As for a special issue on coevolution, TEC did have a special > >> issue on > >> games 9/6 (2005), which did have a few papers about coevolution; yet > >> other papers used coevolution, but were about other things, like > >> evolution of coooperation. So perhaps that 2005 special issue doesn't > >> make a new coevolution special issue inappropriate. I would be > >> able to > >> contribute something to such a special issue, too. I'm open to co- > >> editing, also, conditioned on things like when we plan to do this, > >> etc..
> >> With respect to TCS, if we want "modern" coevolutionary work to gain > >> wider acceptance in the EC community, it might be that TCS is too > >> remote in topic to help with that. One can argue that it will help > >> promote coevolution to a wider audience, however. I guess we need to > >> decide who we want our target audience to be, and the purpose behind > >> selecting that target audience. If we decide to focus on the EC > >> community, to begin with, then perhaps we want an EC journal (i.e., > >> ECJ or TEC). I'd make a similar argument about Natural Computing > >> -- we > >> need to decide who we want our audience to be and why, and then go > >> after the appropriate journal. Also, consider that if propose a > >> special issue for TCS, for example, who of their regular > >> readership is > >> going to submit? To get papers, we're going to have to advertise in > >> the EC community, probably; the set of contributors may be disjoint > >> from the set of regular readers.
> >> To summarize:
> >> 1) other threads and discussions have expressed a frustration that, > >> while a) coevolution is viewed as being important, and b) coevolution > >> events (tracks, tutorials) are consistently well attended, we seem to > >> have a problem A) bringing new people into the field, B) getting new > >> ideas adopted by the community of people using coevolution.
> >> 2) to help with this problem, ideas that have been proposed > >> include a) > >> more tutorials, b) competitions, c) special issues, d) special > >> coevolution-focused workshops outside of CEC and gecco, e) killer > >> apps > >> for the new ideas, and so on.
> >> 3) if we want coevolution (and modern coevolution) ideas to get more > >> widely adopted, and we want to do a special issue, what community of > >> researchers do we want to begin with? Do we want to start with A) > >> people using coevolution, but with older methods? B) EC people > >> generally, who at least have heard of coevolution, may think it is > >> important, and likely know that we are active in coevolution? C) > >> people more outside of EC, who come to coev tabula rasa and don't > >> have > >> biases towards older methods?
> >> Also, to promote the modern coevolution ideas, it seems like we'll > >> have to be the ones to do that; so, a special issue without (or with > >> few) contributions from us will certainly help promote coevolution, > >> but it might not advance the state of acceptance of the newer ideas. > >> Yet, a special issue with mostly contributions from us might be a bit > >> insular.
I agree, we should fish a bit for interest outside of our typical group. What's the best way to go about that? Have you a list of people in mind? We could scan over the last few conference proceedings and email authors of coevolution papers. Or are you wondering about the wider EC audience?
As to many guest editors problem: I actually don't <i>want</i> to guest edit, I'm just willing to help if there's no one else. It sounds like that's not a problem ... so that's one less to worry about on that front.
It's certainly a good idea to get a sense of whether people will want to contribute to a special issue -- that might be helpful in selling the idea to the journal. But, I'm a bit unclear about the purpose of seeing what topics in coevolution others want to publish about. Wouldn't we want to open the special issue to a broad range of possible topics? That is, theory, applications, new algorithms, and so on. It's great to learn about what people are interested in, but I'm unclear how this information affects our decision making process (because I assume we'll have a broad CFP -- maybe that's not what you all have in mind?).
-Sevan
On Aug 22, 2007, at 12:05 PM, R. Paul Wiegand wrote:
> I agree, we should fish a bit for interest outside of our typical > group. What's the best way to go about that? Have you a list of > people in mind? We could scan over the last few conference > proceedings and email authors of coevolution papers. Or are you > wondering about the wider EC audience?
> As to many guest editors problem: I actually don't <i>want</i> to > guest edit, I'm just willing to help if there's no one else. It > sounds like that's not a problem ... so that's one less to worry about > on that front.
Well, we could write a CFP which would only attract papers from people like us. Or, we could write a CFP which is very broad to the point of being vague ("we welcome submissions on the theory and practice of coevolutionary algorithms..."). I expect a very broad CFP would receive a lot of "what I did on my summer vacation" papers, weird theoretical ideas out of left field which are not connected with what's been done, or recapitulations of work from the 90s (it's funny how many people reinvent CCEA and call it a "novel" algorithm). Since we will only realistically have one special issue every few years, we should try to make it as special as possible. :)
That's why I suggested poking around to get a sense for what other people would like to see in a special issue. There have been at least two applications of LAPCA to "real" problems in the last couple years, for instance. That's the sort of middle ground I had in mind.
Personally, I'd like to see stuff like:
-- replication of older results using newer algorithms -- comparisons of several methods on the same (set of) problem(s) -- analysis of cost/performance metrics for algorithms -- anything about representation in coevolutionary algorithms -- how coev scales, particularly questions about building blocks/modularization -- coevolution and robustness
I don't think we'll get submissions focused around topics like this if the CFP is too broad. And I also don't know if that list only reflects my own set of preferences or if it reflects what a lot of people would be interested in seeing in a special issue.
Anthony
On 8/22/07, Sevan G. Ficici <se...@eecs.harvard.edu> wrote:
> It's certainly a good idea to get a sense of whether people will want > to contribute to a special issue -- that might be helpful in selling > the idea to the journal. But, I'm a bit unclear about the purpose of > seeing what topics in coevolution others want to publish about. > Wouldn't we want to open the special issue to a broad range of > possible topics? That is, theory, applications, new algorithms, and > so on. It's great to learn about what people are interested in, but > I'm unclear how this information affects our decision making process > (because I assume we'll have a broad CFP -- maybe that's not what you > all have in mind?).
> -Sevan
> On Aug 22, 2007, at 12:05 PM, R. Paul Wiegand wrote:
> > I agree, we should fish a bit for interest outside of our typical > > group. What's the best way to go about that? Have you a list of > > people in mind? We could scan over the last few conference > > proceedings and email authors of coevolution papers. Or are you > > wondering about the wider EC audience?
> > As to many guest editors problem: I actually don't <i>want</i> to > > guest edit, I'm just willing to help if there's no one else. It > > sounds like that's not a problem ... so that's one less to worry about > > on that front.
|
