It is often possible to move very quickly, and in our hackfests (one was prepended to the symposium) we have shown that it is possible to prototype within a day or two. This creates a new generation of scientist-hackers (I use "hacker" as "A person who enjoys exploring the details of programmable systems and stretching their capabilities" [1]). Several of the authors in this issue would regard themselves as "hackers" and enjoy communicating through software and systems rather than written English. This stretches the boundaries of the possible but also creates tension where the mainstream world cannot react on a hacker timescale and with hacker ethics.

More generally many scientists and information professionals are increasingly frustrated with the conventional means of disseminating science. Most conventional publishers regard scientific articles as "their content" and a very recent article (2011-06-20) from the STM publishers [2] indicates that the publishers believe they have the right to determine how content is, or more often is not, used. As an example most forbid by default indexing, textmining, repurposing, even of factual data to which the scientist has a legitimate subscription. This has an entirely negative effect on information-driven science, preventing even the development of the technology.

While writing this overview and several articles I have become even more convinced that the only way of creating full semantic science is to publish Openly (CC-BY) and to publish completely (i.e. all experimental information (CC0/PDDL)). I believe that most funders now recognise this and are pushing, as hard as they can, to create fully Openly published science. I think this has to come, the question is how long it takes and in what form.

I now believe that in many cases it is unethical to restrict access to publicly funded science. Lessig, in his CERN talk ("Scientific Knowledge Should Not Be Reserved For Academic Elite" [3]), showed that it would cost 500 USD for him to read the top 10 papers relating to his child's condition. These papers are effectively only available to academics in rich universities. A colleague recently told me he had spent a month researching the literature of his child's condition (to critically effective purpose) and we agreed he could only do this because he was a professor at a University. That is one reason I support the Open Knowledge Foundation and its projects to define and obtain Open information (of which Open Bibliography [4] in this issue is typical).

As part of this effort four of us (including authors in this issue) developed the Panton Principles for Open Scientific Data. These principles are simple and, we hope, self-evidently worth pursuing and would lead to a greatly increased substrate for the Scientific Semantic Web. We were therefore delighted when BioMed Central not only enthusiastically adopted the idea but took positive steps to implement this as part of their publication process, for example by labelling data items with the OKF's "OPEN DATA" logo. This is valuable not only in making the data repurposable, but also by promoting the concept - many readers will now be familiar with the logo. BMC have also encouraged authors (and editors) to highlight outstanding examples of data publication (and done me the honour of asking me to present their awards).

It is therefore a real pleasure to work with a publisher who understands my, and my co-authors', intentions and is prepared to work to make them happen. The article explores many new types of publications and BMC have undertaken, as far as technically possible, to implement them as examples of a new generation of publication technologies. I and others have been critical of PDF as a publication format - it destroys semantics and innovation, but we must "eat our own dogfood" [5] and this is shown by several articles. Henry Rzepa creates all his molecules as semantic objects, while in Open Bibliography [4] we use our newly developed BibJSON and ScholarlyHTML to create and publish the article.

The first challenge is cultural; researchers have to be persuaded that Open Data is not only inevitable but also beneficial to their activities. Even when an author is convinced of the value of publishing Open Data, it is usually not trivial to do so. Unlike a manuscript where a static, human-readable, webpage can be posted and served for all time, data are frequently much more complex. They may be very large (petabytes), complex in both semantics and organisation, and even distributed over several sites. In bioscience, it is becoming commoner to see data published as Excel and other spreadsheets but in chemistry (apart from crystallography) the tradition is still to publish supplemental data as PDF, which destroys much of its semantics. One simple and achievable goal of these publications is to convince chemists that publishing in semantic form is "almost" no effort, compared to the effort of producing the data in the first place. If we were able to persuade researchers in computational chemistry simply to deposit their logfiles (usually less than 5 MB), or the Word documents for their syntheses, machines would be able to revolutionise the practice and understanding of computational and experimental chemistry. Open Access (CC-BY) implies (but may not explicitly state) that articles can be repurposed by machine extraction of data items, e.g. by OSCAR.

Many of the problems are cultural and for that reason several of the papers in this issue address the need to change attitudes as much as the technical requirements for the electronic infrastructure. I believe that it is impossible to do modern science unless the key information is completely Open. This applies, for example, to identifier systems, bibliographic data and much factual data. Chemistry, unfortunately in my opinion, has a strong ingrained culture of possession and sale/licensing of data. For this reason, it is often behind other subjects and, in the recent SOAP report [7] chemistry was highlighted as several years behind bioscience in its approach to Openness.

For that reason, some of the things we report are prototypes rather than completely established semantic resources. The biosciences have convinced funders that it is valuable to have completely Open access to sequences, structures, ontologies, etc. In chemistry, most of the freely accessible material has been produced by enthusiasts rather than large funded organisations. Indeed, it is the availability of bioscience resources such as ChEBI which to some extent drive the adoption of Open chemical semantics.

Scientific communities are now common on the web (and even have commercial value) and several of the articles emphasise the role of ad hoc and other communities. The web has the great advantage that anyone can, relatively easily, find those people and organizations who share values and goals, amplifying minority or early-adopter initiatives. Their dynamics are unpredictable and most die, but enough survive to provide world-changing mechanisms.

There is no clear community focus for chemistry overall (though sub sections - such as WATOC (World Association of Theoretical Organic Chemists) may provide one). The main drivers (funding, advancement, commerce) have always been present but the modern era has amplified and often dehumanised them. With growing emphasis on publication to generate the income of learned societies there is a decreasing sense that they act as nuclei for community to grow communal goods.

Because of this, chemistry has almost no public ontologies, and we have a vicious circle. Without ontologies, authors cannot reasonably be expected to create semantic information, and without a clear need for semantic information, the community will not take on the considerable load of creating ontologies. Several of the articles argue that the creation of lightweight dictionaries and other semantic metadata is affordable by the community and I believe that if the communal will is present, then it would be possible through bodies such as IUPAC and others, to create a full semantic infrastructure for much of the current published chemistry.

As an analogy, Mendeleev required access to other scientists' work to produce his classification, as did Pauling, Woodward and Hoffmann. I believe that the current chemical and related literature contains considerable amounts of undiscovered science, and that with 'information telescopes' we can start to discover this.

The development of infrastructure is a lengthy process. The web has, perhaps, given us an optimistic idea of the speed at which new ways of working can be implemented. We are still often governed by Planck's observation ("Science progresses one funeral at a time") and this is equally true for some areas of informatics. Several of the articles reflect the difficulty of catalysing change in what is essentially a mature and therefore conservative discipline.

Henry Rzepa and I were active contributors to the development of XML by running the XML-DEV mailing list (1997). This was a highly successful Open example of true collaboration and for me it culminated in the development of the SAX protocol late that year. XML had been seen as a primarily document- plus typesetting-oriented discipline, but some of us realised its potential for data modelling and transfer, and therefore the need for APIs in XML tools. I nagged continually at the community, and, as a result, Tim Bray, David Megginson and others helped us to develop the SAX protocol, now implemented in every computer on the planet. This protocol was developed in a calendar month and has stood the test of time exceedingly well.

The web can, and, we hope, will, change this. Where you publish should not matter so long as the material is discoverable and the process of reviewing is understood. I believe that the papers in this issue will be read well beyond the cheminformatics community, because their value will be discerned and communicated by methods supplementary to the formal publishing process.

A major challenge in this issue is that the timescales for many of the projects is complex. In many lab experiments (such as chemical synthesis or chemical crystallography) the process is clearly bounded. "make this compound", "check success through crystal structure analysis". Each (normally) has a clear endpoint and can be published as a static document.

In contrast how should we publish software? We use public repositories and these contain a complete record and the current semantic object. If we wish to tell the world about a development we put it on the mailing list. There is no need for a formal publication for those aspects. The motivation is therefore primarily to establish our reputation and there is no simple way to decide when this should be done. JUMBO has had six revisions - should this result in six papers or one or none? (Actually the only JUMBO paper is in 1997 [8]). Six papers would confuse - but after 14 active years it's time for another, I think, which explains the design process. OSCAR3 has its citable publication - a few years back - and we feel it's useful to publish our current ideas, which have more to do with software engineering than new chemical entity recognition.

• Establishment or priority and authorship

• Exposure and preservation of the scientific record

• Communicating the science to one's peers and the wider world

• Allowing the science to be moderated by peers and others ("reviewing").

There is perhaps an additional axis in today's bibliometric-obsessed world: allowing the work to receive an official assessment of merit.

However the publication process is out of sync with the modern web-based world ("Web 2.0") which allows the publication process to encourage and support:

• Collaborative working (as seen in many projects such as Wikipedia, Open StreetMap, and in science, Galaxy Zoo). Here each contribution is often an atom in a much larger cloud and the publication process is continuous rather than discrete. Wikipedia articles are "never finished" though there are some efforts to provide frozen versions. This is a strong theme of this "issue".

• Independence of the source of publication. Given the ability of search engines, and the social networks, to discover anything of value it matters less where something is published. Other than the choice of reviewers the primary issues is whether a piece of information is accessible or limited. History has shown that high quality scholarship on the web will usually surface regardless of where it is published.

• Creation of continuous semantic objects. By recording everything we do, annotating it, and revising it, we can maintain a current semantic publication object at all times, including a revisitable history. This should be the object of scientific publication, not the current PDF.

• The paper (semantic object) as a driver of research. The idea of writing a paper before the research is carried out is valuable and not novel (e.g. George M. Whitesides [10,11]). Here, however, we extend the paper to semantic objects (programs, spreadsheets, molecules, bibliography, etc.).

The long timescales highlight the difficulty of conventional publication. The world knows of these projects through blogs, online resources, user communities and so on, and a conventional learned paper has little value in communicating or preserving. Its prime merit is to achieve a traditional numeric merit for the work, often delayed by several years through the citation mechanism. I believe that it is important to change the values that we use in our assessment of on-going scientific endeavours, and avoid ritual publication.

Some of the articles (Wilbanks [17], Neylon [18]) discuss the philosophy and practice of new models of scientific endeavour and communications. Some of the articles have a retrospective look (CML [13], Zaharevitz [19]) but the fundamental principles are still as important today as when the work was started. A number represent growing points whose development is highly unpredictable. These include the WWMM [16], where the vision of a distributed peer-to-peer knowledge resource has had to wait a decade until it could be implemented. The Quixote project is only months old but takes this vision and has already built an impressive prototype, which I expect to set the model for computationally-based knowledge repositories. These projects rely heavily on community, and this is most clearly shown in the Blue Obelisk movement [20] which aims to, and has largely succeeded in, creating an Open infrastructure for cheminformatics. A major motivation for this has been not just that software and data should be universally available but also that this is the only manner in which science can be reputably validated both by humans and machines. An example of the need for such validation is shown in Henry Rzepa's article [21].

The OpenBibliography project represents a socio-political imperative whose time has come, and for which the technology is appropriate. A year ago the JISC-funded OpenBibliography project could not point to a significant amount of open resources, but in the last year we have helped to catalyse the release of both library data (BL, CUL and several others), and also of scientific bibliography. It is impossible to find Open resources for scientific bibliography but we believe that in a year's time, readers can look back and see this as a key starting point. It is worth noting that the very process of writing this article has generated a great deal of new formalism and tools in Open bibliography, and effectively given major impetus to the BibJSON approach.

The relative stagnation of chemical informatics suggests that change is unlikely to happen from within chemistry. As progress occurs in other areas (retail, bioscience etc.) chemistry may be dragged into the semantic world regardless. If chemists wish to retain control over their own systems they will be wise to start investing in Open semantic environments, because otherwise the rest of the world will do it for them.