Home  |   About  |   Energy  |   Politics  |   Software  |   Music

22 July 2015

Impressions from FOSS4G-Europe 2015

Last week I was at the beautiful city of Como in Italy to attend the European FOSS4G conference. In case you do not know FOSS4G stands for Free and Open Source Software For Geo-Spatial. These days my career pretty much revolves around developing and applying this sort of software in research. I presented a poster promoting the upcoming version 4 of PyWPS.

The conference can only be called a success, fully booked with 400 attendants. This even created some logistic challenges with long food/coffee queues and packed conference rooms. And of course it was great to meet or re-meet the faces behind many useful tools and learn about new software coming down the pipeline.

Here I leave an account of some emerging trends in this field, that may show were FOSS4G is headed.

Open Data


I believe I am not exaggerating if I say that about one third of the communications were somehow related to Open Data. In some cases it was the set up of open data infrastructures, in others open data acquisition and contribution, in others still the usage of open data for decision support, spatial analysis or resource management. Open Data is rising very fast and might very well justify conferences on its own right in the near future.

My impression is that Open Data could actually be more valuable than Open Source software today. A small example: within the Basic Data Programme, the Danish authorities estimate the savings by the consolidated addresses dataset alone to be in the order of 14 M€ - per year. Instead of five different datasets maintained by five different companies/authorities now there is only one. Savings on data interchange between government ministries and agencies are estimated in excess of 50 M€ - again on an yearly basis. All this for an investment of 8 M€.

And there is much more than savings to Open Data. Once the data is public new businesses and services start emerging around it. Open Data is becoming largely regarded as a B2B and B2C enabler.

Volunteered Geographic Information (VGI)


Open Data relies in good measure on contributions from citizens with their smart-phones. There is now even a funny name for it: Volunteered Geographic Information (VGI). OpenStreetMap (OSM) is possibly the best known open dataset entirely volunteered, but alas, it may well be just the beginning. In this conference, beyond the traditional OSM mapping party there were three others that show where VGI is moving next.

The Emotional mapping party introduced EmoMap: a project set to record how folk perceive and evaluate emotionally the surrounding environment. This is a clear path into Social/Psychological mapping.

The Land Cover Validation Game went on for the whole duration of the conference, introducing volunteered assistance to land use/land cover classification. Satellite data may not always be conclusive and require human intervention to correct errors or conflicts - a processes that may well be facilitated if everyone can contribute.

And on indoors, with the i-locate project. With OSM covering pretty much of the known outside world, indoor mapping emerges as a new trend.

A common reference in VGI related communications is the rapid increase in precision and completeness usually witnessed in such datasets.

There's a QGis plug-in for that


The meteoric ascension of QGis to the status of standard desktop GIS GUI has long been consolidated. It was thus not a real surprise to see hardly any other desktop software in the programme (apart from gvSIG). However, there is clearly a new trend setting in: the QGis plug-in.

After a first phase of growth lead by stability and usability, QGis now grows on the ease of plug-in development unleashed by the adoption of the Python programming language. A wide range of users, that may not necessarily have full fledge programming skills, can easily develop tools and utilities that slowly enrich QGis.

In fact, the wide number of communications related to QGis were not on the core of the software but rather on plug-ins. And they come in all colours and shapes: natural hazard management, data extraction and import, graphical interfaces for GRASS models, and more, much more. There are even cases of duplicate plug-ins - pieces of software developed for the same purpose by different folk.

The QGIS Python Plugins Repository counts today more than 500 entries, of which 350 are deemed stable. A virtuous circle has been created in which an ever more powerful software ecosystem attracts ever more users, that contribute ever more tools.

The latest release ofGRASS also supports the Python language and thus a burst of spatial analysis plug-in development is to be expected. The brand new r.green plug-in is a good example.

Sensor Web Enablement (SWE)


This is still a niche in FOSS4G but one where a rising trend can already be devised. In short SWE is a suite of standards issued by the OGC for the acquisition, management and publication of sensor data. Data acquisition has always been one of the major business areas in GIS and it is very likely to continue as such. There were not that many communications on this subject, but enough to foresee a relevant increase in the adoption of some of these standards (SOS in particular).

In this field I was particularly impressed with OpenSensorHub. It proposes a far-reaching SWE architecture, starting with client software at the sensor side, then with an event hub plus a processing engine. The client software also runs on ARM architectures, it may thus be installed on a cell phone or embedded in small single-board computers (e.g. RaspberryPi). OpenSensorHub grows with the coding of sensor drivers, bits of code that enable the integration of observations from a particular sensor. I see great potential in this flexible design.

Git for education


I attended a few communications related to scientific education and a practice becoming common is the storage of teaching contents in a git-enabled web repository. This allows students to contribute directly, correcting problems or suggesting new contents.

It sounds quite obvious, and I am glad to see teachers taking up these novel technologies for the sake of all. But also reminds me how far things evolved since my time, when git was the matter of dreams and GitHub science-fiction. The social side of web technologies is likely to still withhold important advances in the ways we learn and work.

Gender


This is something that was not immediately apparent due to the large attendance but got obvious towards the end of the conference: gender was remarkably balanced. The mapping parties results made it quite clear: in three of the challenges there were more women participating than men and women clearly dominated the prizes.

FOSS4G has hit the mainstream, conferences are not just about a bunch of nerds that code for fun, everyone is in now. The chronic gender imbalance in Computer Science and Information technologies has been diluted by the far reach of these successful tools.

There might be more to this than meets the eye though. FOSS4G conferences are remarkably distressed events, not exactly informal, but everyone seems to be there mostly to enjoy the gathering. Essentially, there is no competition, open source software is primarily a collaborative and constructive processes. This is very different from Scientific conferences - where publication pressure is prevalent - and at the antipodes of Industry conferences - that are mostly about competition.

I will go out on a limb claiming this to be an extraordinary and unreckoned characteristic of open source software. The low testosterone level of this event is the reflex of a social environment and dynamics that by nature is far more comfortable to women. It may well be the case that open source is opening paths for the contribution of women to computer and information technologies that simply were not there before.

Non trends


Finally a note on trends that are not really evolving as I would expect them to be.

In various moments I was left wondering something like: "why didn't they use OGC data service standards for this?" Apparently, these standards are not yet perceived as a general solution to interoperability and all sorts of ugly short-cuts are used instead. This is clearly something left to solve, perhaps through (even) easier to use software. The frequency with which the standards are modified is not helping much either.

3D web-mapping was one of the things I was set out to absorb from this conference. There were a few technical solutions communicated but they seem to be scantly used. There is no single reason for this: client CPU load, lack of demand from end users, lack of third-dimension in base data. All in all web 3D remains a narrow niche market.

The NASA World Wind troupe was very present in the conference and the software seems to be evolving fast, now with ARM support included. It looks like the place to start with 3D.



These are just a few hints on what happened during the three days of conference. Parallel sessions were held in four different rooms and I only managed to be in one at a time.

There will not be a FOOS4G-Europe in 2016, as the international FOSS4G conference takes place in Bonn. See you there.