R in Biodiversity Analysis: rOpenSci for all
Stockholm, Sweden, 24th-25th Jan 2016

Jan 23th 13:55 - 14:15 

    "R Tools For Biodiversity Informatics"

Keyboard shortcuts for presentation viewing mode:

'f' enable fullscreen mode 
'w' toggle widescreen mode
'o' enable overview mode
'h' enable code highlight mode
'p' show presenter notes

• Trust in authorities is relatively high in Sweden, generally perceived to be an open society
• But what about 2017 - the Era of Post-Truth and Fake Facts?
• Domestic Example: Recent Swedish ClimateGate - SMHI / Swedish Meteorological and Hydrological Institute Propaganda Allegations… Is the Sweden where your children grow up getting flooded?
• International Example: The Obama Paper - political propaganda or reproducible scientific facts? Is increased economic growth strongly linked to higer emissions levels of carbon dioxide and other greenhouse gases?

Do we as citizens want to live in a society based on blind trust in authorities and closed sources?

- The case for Reproducible Open Science - to build trust in scientific results through publicly available reproducible peer-reviewed work

• FOSS - Free and Open Source Software - no lock-in, no costly licenses, freedom to create
• Most software on the Internet runs on FOSS stacks and platforms (servers, not clients)
• Jean Tirole wins Nobel Prize for explaining the Economics of Open Source

"Because cats are dog-fooding":

• Huge developer communities increasingly favors open-source and don't want to work with closed source, see Polarizing Technologies on Stack Overflow
• Free support and help offered from the open source communities such as Stack Overflow, GitHub, Docker Hub

Purchasing behaviour and Budgets follow their own rules:

• "Perpeutated Budgets" - budgets and cost levels tend to "stay the same" over years - consider the "end of year" purchasing frenzy phenomenon, aimed at keeping the same budget levels for the next year
• "Siloed Budgets" - if there was one budget instead of several budget silo compartments: then savings on licenses and hardware -> investments in people and processes
• "Non-Humane Budgets" - If (People > Machines) then don't purchase costly reoccurring closed source software licenses when FOSS alernatives exist - it is a waste of tax payers' money that could've instead been spent on improving services

• International trends - DevOps, Data Science, Big Data, Machine Learning - mostly based on FOSS tools
• Data Science using R, Python, Julia, Scala
• Even Micosoft goes from closed source to recently supporting Linux Foundation… also recently buying Revolution Analytics now "Microsoft R server", for providing a fast R runtime commercially
• Systems Integration of best-of-breed FOSS open source software components using Docker - a FOSS tool that makes it possible to build, run and ship (almost) any software (almost) anywhere through "light-weight containers"

• Not one single centralized local system developed in splendid isolation and expected to spread across the world, instead a decentralized web of latest web-based open source system components - developed openly in international collaboration, up-to-date and mirroring the latest advances within Free and Open Source software for Data Science
• Enabling cost-effective reproducible research work, removes any unnecessary non-free software licenses - respecting the tax payer financed funding model - avoids "double taxation" through costly commercial business models, instead providing public services, publishing results openly, providing transparency for both sources and results

• R for Reproducible Open Research using ROpenSci - lots of packages available immediately for use in biodiversity analysis and other areas - providing a web-based frontend for experts authoring reproducible research - based on web-variant of RStudio R IDE, customized - and also simple-to-use web UIs for non-experts
• Data services from open source software based data aggregators such as GBIF and Atlas of Living Australia and any other Web APIs or open data sources

User interfaces for the Web

• Do you want to provide web user interfaces usable by non-experts on any platform? Create a Shiny web app and bundle it into your R package.

Data for the Web

• Do you want to turn your R functions into a web service? Regarding the "serverless computing" buzzword: see this impressive R package OpenCPU - Producing and reproducing results allowing you to turn any R function into a web service and bundle it into your R package.