(This proposal of mine was first published at per-cloud.com in February 2017. It is now published again here, on invitation by M. Bauwens. For more context and details, I strongly suggest to also read, before or after this proposal, the posts from my own blog linked at the bottom)
important update, 2018/02/06: a new version of the proposal, completely rewritten to take into accounts recent developments and feedback, is HERE.
Preface
A percloud (permanent/personal cloud) is my own vision of a “REALLY usable, all-in-one alternative to Facebook, Gmail, Flickr, Dropbox…”.
I made the first percloud proposal in 2013. Very soon, however, I “froze it”, for lack of time and resources, and did not do any real work about it, for reasons I have explained in detail elsewhere. Then, at the beginning of 2017, several things happened, including but not limited to:
- The percloud proposal was mentioned here, causing several people to contact me to discuss the idea in detail, showing to me that it may still have some value
- In parallel, the Free Knowledge Institute, of which I am a Board Member, had started some work in Barcelona on a collaborative/community cloud platform
- Other groups have started to work on similar platforms on their own, and invited me to participate
- I did some homework to catch up with the “latest” developments in this space, and discovered that things look much better than they did in 2013
What I mean with the last bullet is that, thanks to projects like Sandstorm, Cloudron and several others, building what I call a “percloud” should, indeed, be easier than in 2013. “Easier” does not mean “easy” though, and I have realized several things.
First, integrating and polishing the several software components, until they are actually usable by non-geeks is still nothing one could do on his spare time (not me for sure, anyway). Second, personal clouds will be easily adopted by non-geeks ONLY if they are offered as a managed service: this means there must be web hosting providers that offer really turn-key perclouds.
Third, a real pilot/field trial of the percloud is needed. Because on one hand, we need many, ordinary Internet users to use the package, and tell us geeks if it works for them or not. On the other, we need to give wen hosting providers some real world usage data of these personal clouds, so they can figure out how much it would cost to offer them as a service.
Taken together, all these things have lead me to put together the proposal below.
Important: as I said, I’m already discussing similar cloud platforms with several groups. But I do not see this proposal in competition with the others. This is all Free as in Freedom software, and the more is shared and reused, the better! Much of what is proposed below may be directly reused in those projects, or similar ones, if not co-developed together.
Now, please look at the proposal, share it as much as you can, give feedback and, since this page may be updated often in the next weeks, follow me on Twitter to know when that happens. THANKS!
Percloud proposal, 2017 edition
Percloud definition and features
Purpose: personal, permanent, basic, online web presence and communication, that does replace {facebook+gmail+dropbox} today. Very little or nothing more. The target user is the average user of facebook, gmail, instagram, dropbox, google drive and similar services, who seldom, if ever, visits the rest of the Web. The goal is to make it possible to these people to get outside today’s walled gardens, as soon as possible. Once that happens, it will be much easier to move the same people to more advanced platforms. Advanced users for which this service is too little/too limited still need something like this for all their own non-geek contacts, if they want their communications to stay private.
Services offered
(regardless of which software implements them…)
ONLY the very basic ones, that everybody would surely need, e.g.:
email, blogging, calendar and address book, basic social networking, online bookmarks, save web pages to read them later, online file storage (personal files, pictures galleries).
common essential features of all services
- Inclusiveness (“equal opportunities cloud”?): the percloud must be an accessible service even for the many people who, these days, have a smartphone, but NOT broadband, fixed/safe residence, reliable electricity… (think students, but also refugees, migrants, homeless…)
- Available as a service (PEAAS, Percloud As A Service): even many target users who could run their own hardware server at home will prefer the convenience of not having to worry about any additional device
- Federation (where applicable): that is automatic notification of relevant events among different perclouds (e.g.: user A uploads a public picture or status update, her contacts see a notification about it in THEIR clouds, and can comment it, and be notified of each other comments…
- Social Connectivity with Facebook, Twitter, Gplus… and interface to online storage services like e.g. Dropbox… Google Drive
- here “social connectivity” means a) possibility to automatically publish a status update also to Facebook, Twitter etc… and b) fetching Twitter timelines or Facebook notifications with systems like these http://www.techrepublic.com/blog/linux-and-open-source/how-to-access-facebook-from-the-command-line/ and showing them INSIDE the percloud interface. Interface to storage like Dropbox is the same as in owncloud/nextcloud
- PURPOSE: the target users will NOT move to this if it means burning bridges with their friends already on Facebook, etc.. No way.
- As little choices and configuration options as possible: “You can have any color as long as it’s black.” For the target users, this is a feature, not a limitation
- ARCHITECTURE:
- 100% server-based. A permanent online home, and important data, cannot live on a smartphone or laptop, which may be stolen, or run out of charge every moment.
- in practice: micro-vps: bare-bone Linux with all and only the services listed above, that can run on raspberry pi, normal PC, virtual hosting in a data center… also because of…
- HIGH performance, i.e. RAM and CPU requirements as low as possible. Reasons: a) a really personal cloud of this type has very low needs anyway; b) only if it is possible to host MANY of these VPSes on one physical server it is viable to offer this as a managed service. Ideally hosting one percloud in a data center should not cost more than a few USD/month
- REAL, almost turn-key portability from server to server. This includes automatic set up, at creation time, of own domain name, e.g. “marco-percloud.com”, so that even when changing server all connections with the rest of the world stay the same. There is no real freedom or data ownership if one cannot move her own data from one physical place to another without losses or disruptions of communications.
- MANAGEMENT:
- complete separation of user interface and admin interface
- remote administration also possible via command line
- reason of 2 previous requirements:
- PAAS providers can only afford to offer the service if administration (creation, software updates etc…) of many perclouds can be fully automatized *if admin is a separate account, it is possible to manage or give perclouds as “gifts” to minors, senior citizens, or everybody else who would like to use a percloud, but cannot or does not want to manage it themselves
- automatic daily backup to other server
- USER INTERFACE:
- Accessible from any modern Web browser, from any device (i.e. NOT locked to phone numbers or anything like that)
- But REALLY usable on smartphones! Many target users only use their smartphone for any digital service, and will not accept something that THEY find hard to use on their preferred device
Possible base software platform:
- barebone GNU/Linux distribution (debian? TBD) +
- SUBSET of the cloudron.io environment with ONLY these applications and 3rd party additions
- owncloud or nextcloud (online file storage, picture galleries, office suite too?)
- Lychee (photo management)
- SoGo or Radicale? calendar and contacts
- Wallabag (“Read later”/personal web archive service
- RocketChat
- Rainloop for webmail (but see “3d party SW” too, below)
- Ghost for blogging (static blogging with markdown editor is simpler and safer than wordpress, and should require less RAM. See performance above)
- Wekan or Kanboard?
- Piwik
- 3rd party software, i.e. packages not in cloudron today, but that would be nice to have IMO:
- movim.eu (social networking like that is crucial, IMO)
- mailpile for webmail?
- shaarli (nice online bookmark application)
- SSL key management for web and email servers
- GPG signatures management
Note on interface integration and “real time interactivity”
The several components of a personal cloud as proposed here would share user authentication, and communicate with each other, as smoothly as possible. However, they cannot have a completely homogeneous look and feel as, say, the several features of a Facebook account. Such an integration is simply outside the scope of this proposal, because the only (but crucial) purpose of the percloud is to test and offer something actually usable, as soon as possible: see the “we need it SOON” part of this post, which is even valid now than it was in 2014, to know why.
As far as “real time interactivity” goes, the percloud must offer federation, that is let “friends” who own different perclouds see what each other has published, comment it, get notification, chat, and so on. However, percloud-based social networking does not even try to achieve the same numbers and levels of interactions and notifications of Facebook or similar platforms. This is a feature, not a bug. Facebook bombards people with real time notifications (“Jim tagged you”, 3 years ago you posted this”…) because it exists to… make people stay as much as possible inside Facebook. A percloud, instead, exists to let you interact with your contacts when you need or feel like it. It does not need to be so invasive and stressful.
Looking for sponsors
The contacts and discussions I had at the beginning of 2017 convinced me that a percloud available as soon as possible may still have a lot of value. The same activities also showed me that it should be done quite differently than what I imagined 4 years ago.
In order to build a percloud and test it “in the field”, together with the cloudron developers, it is necessary to have sponsors for: * adding the missing parts * integrating and documenting everything * CRUCIAL: deploy and manage a “large” scale field test/pilot in which e.g. 1000 people are given one percloud for free, for 12 months, in exchange of giving feedback on usability, etc… and allowing basic monitoring of percloud usage (e.g. number of posts and visitors per month, etc). Without this, i.e. without knowing for sure how the actual target users react to the percloud, we cannot make it succeed
As far as hosting goes, the test perclouds may be hosted on lightsail or similar platforms. But it would be great if community-oriented hosting or connectivity providers like guifi.net or mayfirst.org wanted to participate. If you know of any organization or group of organizations who may be interested in sponsoring such an activity, please let me know.
Further suggested reading, added on January 30th, 2018
My proposal to integrate the percloud with the Eelo operating system for smartphones
- My posts on Mastodon, online identity and the need to put INDIVIDUALS before “platforms“
Photo by kndynt2099
To solve hosting and scalability, can it be built as decentralised network/service, say, on ethereum?
Servers are a mess to handle. Ideally, client software (file sync manager, or even browser addon) would act as servers for other users during idle time, using allocated amount of cpu and bandwidth.
Hello Dudenas and.. I’m confused. Why do you say so?
Servers are a mess to handle? Haven’t you read the “percloud as a service” part? Something like that is WAAAAAAAAAAAY less more complicated to implement than anything on ethereum. Or anything like “client software acting as servers for other users”. THAT would be a mess to handle, way more than the equivalent of one self-sufficient Linux container that you can run on your own computer, or host in any datacenter.
FIY: I have just completely rewritten the percloud proposal, taking into account the current general landscape, and all the feedback I’ve received in the last weeks:
http://per-cloud.com/percloud-proposal/
Please read, comment and share it, thanks
FIY, the latest version of the percloud proposal is now also available as a shorter slideshow: https://www.slideshare.net/mfioretti/some-thoughts-and-a-proposal-about-personal-clouds-to-replace-corporatecontrolledplatforms