Paperpile: A new take on reference management

Presentation at the Second national gathering of Swiss health librarians, Bern September 8, 2014

Transcript

1. Paperpile A new take on reference management Second national gathering of Swiss health librarians, Bern September 8, 2014 Stefan Washietl, Founder
2. Outline 1. Background information 2. Demonstration 3. Discussion
3. The Paperpile approach 1. Product centered 2. Built for the “Gmail generation” 3. Opinionated 4. Problem and detail focused 5. Design driven 6. Support driven 7. Constantly evaluated
4. Product centered Correct and reliable Intuitive and efficient Enjoyable • Paperpile is not built on an abstract company vision to “revolutionize science” or “disrupt publishing” • Paperpile’s goal is to build the best possible product in its category • Based on fundamental engineering and software design principles
5. Built for the “Gmail generation” • We observed typical productivity workflows among academics: ◦ Google Calendar as shared group calendar ◦ Gmail as central communication hub. Institute emails are often forwarded to Gmail. ◦ Google Docs is used to collaborate on documents ◦ 60-70% of academics use Chrome as their primary browser • 70% of universities in the US are Google Apps for Education institutions • 100% of universities are de facto Google Apps institutions also in Europe
6. An ever-changing computing landscape • Current reference manager built for the Desktop, web is an afterthought • Paperpile built on the idea “How would a reference manager for Gmail users look like?” 2007 2006 2008 Papers Springer Mendeley Elsevier Zotero George Mason University 2001 1989 EndNote Thomson Reuters RefWorks ProQuest 2011 2012 Chrome OS Google Drive First version of Chrome Chrome most popular browser world-wide
7. Opinionated • Paperpile development is driven by a clear product strategy not by the urge to fill all dots in a feature matrix. • Users get excited by a product that shares their vision and not by a product which is the least common denominator of all possible strategies. • Paperpile is not for everyone and that’s OK. Choice is good: ◦ Well integrated Desktop app for Mac ⇒ Papers ◦ Free, cross-platform app for the Desktop ⇒ Mendeley ◦ Open source, natively built into Firefox ⇒ Zotero
8. Problem and detail focused How we select and implement features ◦ Must be a real problem not a “nice to have” feature ◦ Don’t get distracted how other products solve the problem, solve it from scratch ◦ Iterate until the most efficient and intuitive solution is found ◦ If making things easier for the user makes our life harder we accept this. ◦ Go the extra mile to get the edge cases right
9. Example: Sharing papers with colleagues • People send PDFs via e-mails all the time • Practical issues: ◦ Sending multiple (potentially large) PDFs to multiple people via e- mail is inefficient ◦ Want to share context (e.g. original source, meta-data, supplementary files) not just the PDF ◦ Opening e-mail attachments can be clumsy e.g. during a Skype call ◦ Application specific solutions can work well but requiring everyone to use a specific reference manager is not an option
10. Example: Sharing papers with colleagues • Private sharing links • Meta-data, fulltext PDF and supplementary files are available immediately for everyone with the link and a web browser (no login required) • Link can be shared via e-mail or any other way (e.g. chat during Skype call) • Contacts can be imported to Paperpile so you share with your actual colleagues not just other Paperpile users
12. Is PDF Is Website Is database entry Rule based crawling for meta-data Database lookup Find original website Combine results for best possible data Rule based crawling for PDF Download PDF Local cache for immediate viewing Google Drive Persistent storage and universal access Analyze PDF for meta-data Meta-data PDF Example: One click data import
13. Publisher coverage
14. Fallback options • Design for failure • Think about what users would do and automate the workflows as much as possible
15. Design driven • Design is core to product success. • Not an evolution of other products. Designed from scratch. • We tried to learn from the best web applications not from other reference managers. • Core principle: say NO to everything, YES needs a good reason.
16. Design principles • Consider all levels of design ◦ User interface design (UI) ◦ User experience design (UX) ◦ Aesthetics • Challenges ◦ Present dense data readable and concisely ◦ Present complex workflows in an understandable and intuitive form ◦ Actions will fail (e.g. network problems, user does not have access to a PDF) and have great frustration potential • Solutions ◦ Clear hierarchy of importance of data and functions ◦ Context: only show what’s important now ◦ Design for failure. Explain failures and offer solutions
17. Support driven • We don’t outsource support. No layer between user feedback/complaints and developers. • Direct motivation for developers to improve the product. • Within the first year we addressed hundreds of edge cases which can only be noticed during real-life use. • We take every request seriously.
18. Constantly evaluated • Central question: Are we building something people really want? • If so, it can be a rewarding experience for us as company and our customers who use our software every day. • If not, we have failed. We would not like anyone to use our software because: ◦ They don’t know any better alternatives ◦ It’s “good enough” and free ◦ Their university told them it’s great ◦ We told them it’s great
19. Hypothesis before launch: academics are not happy with their reference managers Hmmm... Zotero got me through my PhD, and the import button is fantastic BUT the application always felt a little clunky. Having to use bib, dropbox and rsync to get everything sorted and available on all devices Oh god! Don't get me started! I want to be able to create and store all of my references for a Google Doc and access them everywhere, there's nothing that integrates seamlessly yet. I've been waiting FOREVER I use Mendeley, but the interface is poor and cannot fetch pdf or references. I hate having to download all the articles myself, and then separately download the citations. Help me save time!! It's all a mess. Wish I could do GDocs and not have to put the doc through 4 postprocessing steps to get the refs in. No good Linux clients! Web- based solution would be excellent for those of us that use many platforms. Not being able to quickly save and tag a reference. I get lazy, don't do it properly, and I lose it. Keeping everything in sync across machines/platforms Lack of cross-platform compatibility and poor BibTeX support. Collected more than 500 individual opinions before we even started
20. Reactions after launch There are enough hardships in academic life without having to fight with references and citations, and Paperpile is the first service I’ve used that makes this truly painless. PC World (March 2014)
21. Don’t trust what they say, observe what they do Users rely on Paperpile for mission critical tasks: Writing papers and grants Users switch from free products to Paperpile which costs $36/year You have a great product, one of those I'm happy to pay for! — A user from Mexico Paperpile is the only web service I pay for — A user from the UK “Power users” and notoriously busy users (clinicians, young assistant professors) switch to Paperpile I've spent a long time trying and rejecting PDF managers (Zotero, Mendeley, Readcube, Colwiz, Qiqqa) but this is the first one that I feel could work for me! — Early beta tester who has been an active user since 1 year now Paperpile lets me work more efficiently and concentrate on what’s important. — Anesthesiologist and critical care physician, Norway
22. Demo
23. Bringing Paperpile to Universities • We are launching site-wide licenses for universities in the coming weeks. • Next step in our ongoing learning process. • Our goals: ◦ Provide students and faculty with uncomplicated access to Paperpile ◦ No technical or administrative overhead, i.e. full self-service for end users. ◦ Competitive pricing
24. Thank you! Try Paperpile: paperpile.com Twitter: @pprpile Contact: support@paperpile.com