My recent paper on the potential value of open source software in Hydroinformatics has stimulated another response (in addition, that is, to those published in the Journal from Profs. Mike Abbott and Jean Cunge). I will respond to Ari Jolma's comments here quoting, with permission, from his email.
I read with great interest your paper in J. Hydroinformatics. There surely is a need for free software and open and useful standards in Hydroinformatics.
The word "useful" in this sentence, apparently so innocuous, is critical, I think. I managed to resist all but the briefest comment on this subject in my thesis, because it was clearly off topic, but that brief comment I did feel compelled to make. Interoperability, Not Standards is Clay Shirky's mantra. Shirky was talking about premature standardisation in the context of Peer to Peer software. His conclusions seem valid for the hydroinformatics world too, however.
Indeed, the coupling of standards and interoperability is the default for any widely dispersed technology. However, there is one critical period where interoperability is not synonymous with standardization, and that is in the earliest phases of work, when it is not entirely clear what, if anything, should be standardized.
I believe this is the stage we are at; we understand that there is a need to be able to build models which integrate features currently only available separately. However we don't have any kind of theory about the nature of the integration work to be undertaken. The HarmonIT project is likely to produce many insights into just what it is that is really required, and is likely to support the creation of software which would be impossible now, but to set out to establish a standard from a first pass at the problem seems ill advised. [Aside: the comments I made on version 0.6 of the OpenMI Architecture reports are online.].
Back to Ari's comments.
Abbott's comment is typically cryptic and Cunge's is strangely almost hostile -- at least to the "you can find everything on the net" attitude, he is barking at something I can't quite see.
Neither response came as a great surprise (excepting that to get a response at all came as a surprise), although I had never communicated with either before.
Abbott's discussion fits in a theme which he has been developing recently, which exposes a revolutionary streak which might be more expected of someone of, well, my age (but which, for being found there, would be far less compelling). His comments allowed me the opportunity of making some political comment which I had deliberately avoided in the original paper (with qualms, I might add; I hope I never have to defend that decision to Richard Stallman, who objects to the open source label because it specifically denies the politics of free-as-in-speech) and stimulated, as all dicussion should, some further thoughts.
Cunge's hostility struck me, too, but his objections I am sure will be often repeated. His observation regarding the existence of different types of software is an important one, though the particular separation of OS software and Hydroinformatics software fails to convince. As for finding "everything on the net", I make the point in the response that the problem of the blind acceptance of results from modelling tools (a problem worth barking at) is orthongonal to the means of production and distribution of the software. Indeed the social networks established in online communities seem to me to provide the opportunity for greater protection from this than the simple fact that the software comes from an established company.
The issue of blind acceptance is closely connected with the points I made the other day about teaching.
Other recent posts here will make it clear that I am very much "anticipating future development of something else" (something other than current modelling software). The content of my thesis will further make it clear that there are significant problems to be overcome in the way hydroinformatics software is conceptualised needs to change. Hydroinformatics software has adopted the application mind set, where lots of little island of functionality are created which cannot easily be brought together into an integrated whole; this I believe is at the core of our failure to move forward from the fourth generation of modelling to the long anticipated fifth.
I argue in my thesis that our approach to hydroinformatics software must change in two main ways.
- We must stop thinking about "makers" and "users" (of software, originally of modelling software) as two well defined and disjoint sets. I have submitted an abstract for a paper on this subject to Hydroinformatics 2004. The "users" of the products of companies such as DHI are often in fact "makers" of tools for others. The consultant preparing a model for a government agency of a particular river reach in MIKE-11, for example, can be seen to be a tool maker just as much as a tool user. This is important because tool makers make very different demands on their tools, and these demands are, I believe, not being met.
- We must stop thinking about applications. This will, I think, become a recurring theme here on my weblog [1, 2].
In fact these are two sides of the same issue. The way to provide tools for these tool maker/users, which exist in great number even if they are not acknowledged (and may not even recognise themselves as such) is by building platforms on which they can build their tools. The resulting scenario, in which many people develop smaller pieces of a larger whole, also makes the value of the open source development model easier to realise.
The idea of "open source" become somewhat sexy in hydroinformatics few years ago but I have not yet seen much real activity towards that. The community is missing.
The HarmonIT project at least intends to release reference software as open source; the impact of this remains to be seen (see above). I'm intrigued about this sexiness; when I embarked on the paper I had heard little or nothing of it. I fear this may be a reflection of my narrow field of view rather than of reality. In particular I need to start getting to more conferences, such as
I have not attended the hydroinformatics conference since '99, I've been more to the ModSim, iEMSs and ISESS conferences.
Community is key. But I think it is fair to say that community follows software in the open source world. I have, of course, contributed to the noise by writing the paper under discussion, but really the way to move these ideas forward is to write, and release, some software, and to start trying to build a community around that. I hope I will be able to do that soon. I intended to release the software developed in the course of my PhD, but really it served as no more than a tool for exploring and developing ideas. The work required to clean it up for release is out of proportion to its potential value.
There are an active free software communities in GIS (freegis.org) and remote sensing (remotesensing.org) which are quite close to water engineering. What is the difference? Maybe people doing GIS and remote sensing are a bit closer to the computer & programming. In water sector it has always been a bit like necessary evil -- that's my impression at least.
Not something I've thought about before so specifically. The generally low level of expertise in computing among engineers has concerned me for a long time. I worked for Arup between school and University, and even then I was able to do things, simple things, that people seemed to find indistinguishable from magic. I came across something the other day about differences between incompetent, competent, and expert. The idea was that an expert is able to work around or without the specific tools they are used to, where someone competent knows how to apply those tools well. I can't find the link now, annoyingly.
Yes, you're probably right, although ESRI are doing the damndest to turn GIS into an application. In remote sensing it is certainly the case that the people doing it are closer to computing.
The "productisation" of hydraulic modelling, a celebrated part of the arrival of the fourth generation of modelling tools, pushed these tools out into places where their new users had not started out by gaining a thorough understanding of computing. ESRI, I think, are engaged on the same process. Remote sensing is still an experts' game.
So what if the "financial environment" is against free software in hydro (or other) informatics? As long (or if and when) we get a community who shares the free software goal that should not be our problem. I see some dark clouds which are:
- software patents
- agreements which take the copyrights away from employees
Software patents do indeed represent a threat, although the tendency for new developments to move from academia might reduce the impact of this (for a time, at least). Copyright ownership agreements have been around for a long time, and are not insurmountable barriers. We have a lot of educating to do.
Comments