Interview Ed

18 Nov 2019

Expert

Summary

Ed is an experienced developer working on archival systems for cultural heritage institutions and museums and very involved in working with P2P protocols and file stores. While working heavily in the sector Ed is very skeptical about user uptake for P2P technologies and think efforts should be instead concentrated on the academic and scientific sectors.

Key findings:

• Average users don’t care about URLs and find them confusing as their structure was never meant to be human-readable necessarily
• Browsers, besides Chrome, are very bad at explaining what is going on with addresses and security
• URLs and the address bar should probably stay just the way it is as users hardly use them but still needs to be there
• The average person doesn’t care where the data actually is
• P2P has very high barrier to entry as many average users don’t understand the advantages and why it may be worthwhile for them
• P2P efforts should be using academic and scientific publishing and archiving as a use case as immutability and permanence are vital

Interview notes

• Works in cultural heritage and archives for 7 years, largely for the V&A and British Library - working especially around GLAM in the sector
• He’s been following P2P especially in DAT and IPFS
• In particular he’s been working on connecting BIIIF and interoperable data, for instance combining images and other media with P2P
• UV.pub - Universal Viewer (based on DAT)
• Jim Pick (ProtocolLabs) used in a demo
• His interest is in getting BIIIF in use and making it as easy as possible to have people use this stuff
• At Mozfest got it work on Glitch and building examples - for instance with making zines which you can remix with DAT and IPFS versions
• He’s interested in helping expand URL specifications
• It is quite difficult all the logic around files and folders, for instance P2P folders and URLs in JSON files
• Even the lowest barriers to entry with P2P like Omega are very hard and wanted an easy way to publish so did the Nomad Project
• File connection with folders have to name in a certain way and what this does is basically make a static site generator
• Trying to lower the bar of entry for this data and iterating quicker and tapping into hobbyists more now
• Adoption of protocols is very difficult
• He knows the developer of Beaker browser and supports P2P in general but concerned about usability
• He spoke with Tim Berners Lee once about URLs and not being able to share or even read
• Nobody does the typing in the URL
• Why we have domains
• Nobody understands content hashes, they want condense and shareable links
• Gave up on IPFS and moved to DAT because of IPNS and immutability
• If you had a cat.jpg and if you had a hash of it, if it changes than you would get a new hash and you would know it
• Functional programming has a UX problem
• We need to be building side effects into languages
• Content hashes and immutability works like branches in git where it changes over time - the address is stable and permanence is critical
• DAT doesn’t have content addressing and puts everything like DNS in a text record
• How do you fix URLs? Or people’s understanding of them? Solution should be to keep on using them like people do, which is hardly.
• The average person doesn’t care where the data is and the promise of P2P is flawed for most users, except for museums and libraries. Nobody wants to keep all their data on a Raspberry Pi or something like that. It’s like keeping money under a mattress instead of a bank
• He doesn’t want to host his own version of Google Docs. As long as he has the rights to his data, then someone else can host it
• Beaker is a great idea but nobody is going to use it
• People don’t care about the data, just that they have rights to it, i.e. Google has a checklist who they need if you want to user their services such as maps. Google is also really good at providing additional information and showing how it all works and letting you agree to access and when.
• Domains are like brands, you trust them
• Communicating things like with the padlock isn’t that great as most people don’t know what SSL even is
• Beaker have a thing where they show how a website is being seeded
• Napster was great because it would show you how many people you could or would be downloading from - that sort of info makes sense
• Sharing cultural heritage resources should be open and they should be using P2P - also with scientific and academic research
• FAIR (Free, Accessible,….) is a great concept with sharing of files and he’s working on something with Duke University and using Universal Viewer with 3d models, sort of like a Napster for scientific data sets. In this case the immutability of content addressing it would be tamperproof and good for citation.
• Libraries and archives are always asking now if they need blockchain
• He wants to provide on-ramps to create BIIIF data and show over HTTP and P2P if you have that enabled or want it
• P2P suffers from the MetaMask extension problem, nobody wants to install it and should be transparent
• He’s interested in keeping in touch and collaborating