OpenSocial Specification | Hacker News Reader

OpenSocial Specification

github.com

107 pointsecliptik30 comments

So many thoughts about OpenSocial and the reference implementation, Shindig. I have it thank for my time at hi5, LinkedIn and then Google.

Some little known facts about OpenSocial

- Hangouts Apps (remember those?) were based on OpenSocial containers.

- OpenSocial powered the LinkedIn Apps Platform and Labs for a number of years. The team built Rails and Node apps and deployed on Joyent.

- Eric Schmidt gave a pep talk to the working group pre-launch and mentioned about how open always wins in the end...

- MySpace was concerned about the attack surface of 3p apps running in iframes. They toyed with the idea of requiring a webkit browser plugin to run apps (!). It did lead to Caja* as a project. https://en.wikipedia.org/wiki/Caja_project

- The work on OpenSocial led in small part to the Activity Streams spec which led to ActivityPub and thus the latest Fediverse protocols. I like to think of OpenSocial as dead, but a good organ donor.

Fun times...

Amusingly a lot of the concerns about embedding applications in other web applications have finally been mostly covered by the browsers adding the capability to sandbox iframes. We used Caja at Yahoo but it was a nightmare at the time.

https://www.html5rocks.com/en/tutorials/security/sandboxed-i...

Hey Paul/Ben-

Always enjoyed working with you both, and sitting on panels teaching. Small world, I'm back working with Dave Recordon again.

I'd love for all the hard work that was left on the table to get resurrected, I continued to work in Caja-style JS sandboxing for a few years after OpenSocial.

Cheers, :max

Hey Paul, long time - I enjoyed working with you on OpenSocial stuff back in the day when I was at MySpace.

I remember working on 5 different apps on the LinkedIn homepage, was great working with you Paul!

That's some internet history right there. I actually used open social back in the days. Of course post Cambridge Analytica people now think that open systems and data portability are bad :(

Similarly, remember the conventional wisdom in the 90's of "To stay safe, don't use your real name online"? But today real-name policies are in place apparently to "keep us safe online".

I was just thinking about this the other day. With Facebook popularity in decline, what major social media still insists on using real names?

To be fair, I don't read their ToS very closely, so maybe they all still require it.

Twitter technically doesn't, but they're close with their all-but-mandatory phone number confirmation, given that many countries now require government ID to get a SIM card.

If the internet is to be viewed as a utility, anonymity is going to be a hard question to address. Today, a water bill is used for identification in getting a license. In the future, government agencies might just have their own tracking cookies.

It's a win-win: you'll get through the DMV line even quicker! /s

Well, your internet already is billed like a utility, but just like how there is no utility tracking utilization by person in a household, I would expect no such requirement for data. That doesn’t mean ISPs and other players won’t try.

It's not a fool-proof link though. Tons of public access points, work/school networks, and multi-tenant home networks.

I actually don't think people have a consistent position here. For most, whether or not an open system is good or bad is largely dependent on how it affects their political group...

I wrote one of the first OpenSocial wrappers so that making apps across various social networks was faster. Not all social networks perfectly supported the APIs as spec'd and had quirks:

https://code.google.com/archive/p/opensocial-framework/downl...

I made roughly 3 applications using it so that I could be faster than all the other developers building apps. Lots of people downloaded the framework when it was opened and some even paid for private copies of it early on.

By the dates i was doing something very similar at large Co around the same time. We never launched the public product, but my container was extensively used internally for development of "apps" for Orkut et al over the years.

Fought for a while to release the code, but in the end assumed the work i was having to get code opensourced was too little as everyone already had theirs by the time (apache-shinding era then). In my view mine only added good cd/ci practices over what i saw elsewhere.

There have been a million reasons for years to support decentralized communication platforms, but it only gets a burst of attention when a fascist / race-nationalist movement gets de-platformed? I find that more than a little depressing. It says not so flattering things about this community if this is what you folks care about enough to notice what's been obvious for so long.

Big social networks have been mass surveillance and manipulation machines for years. The cloud and locked-down mobile platforms have been dangerous tools of mass control and market manipulation for years. None of these platforms have ever been free, open, or unbiased. Absolutely nothing changed in the past 48 hours.

If you don't like what just happened, it's your fault for not caring for 10+ years. (Unless you are one of the few who have been paying attention...)

> There have been a million reasons for years to support decentralized communication platforms, but it only gets a burst of attention when a fascist / race-nationalist movement gets de-platformed? I find that more than a little depressing. It says not so flattering things about this community if this is what you folks care about enough to notice what's been obvious for so long.

This is a pretty reductionist way of looking at a post like this. These events have opened up a lot of interesting points:

- The practice of websites "soft deleting" content

- Reckless data collection

- The ease at which managed infrastructure can be revoked by vendors, without warning

- The potential for harm that bad actors can have on a centralized network

- The limits of curation / moderation

- The degrees of separation between companies that manage infrastructure, manage application code, and the people that manage communities and the power dynamics in between

- The roles of narratives, misinformation, and how it can be both monopolized, weaponized, and used to radicalize

Thinking about these things critically and iteratively does not make anyone a far-right apologist. Decentralization has been discussed on this website off and on for as long as I've been here. It's not always relevant, it's a problem with lots of caveats, and it will take rediscussion over time to find a solution that works well for everyone.

Still, I can't help thinking that it took opening this particular Pandora's Box to have this discussion. The last point in your response might not have been visible until the Capitol attack, but the rest was visible to anyone sufficiently pessimistic about the modern tech/social/legislative landscape.

Maybe you've missed the other decentralization discussions. I came here because of them, although I was (and still more am) concerned about decentralizing code hosting.

I am aware that this was a discussion on the peripheries like the small web/privacy advocate communities. What I'm surprised is that this wasn't discussed earlier in the technical _mainstream_, which is a distinctly different audience.

Ah, that makes more sense. I missed that part, my bad.

I 100% agree with you.

I submitted this because after seeing all these alternative social media, messaging, etc links the last few days I remembered the dreams of the aughties and thought OpenSocial deserved some attention.

The other link I submitted today was for WASTE - the Nullsoft anonymous encrypted P2P chat client from the early aughts, but that one didn't seem to gain traction.

Many of these ideas have been around for a long time in a much more rational form and should be re-visited and built upon instead all these fly-by-night knee-jerk response apps.

We built a really cool enterprise social network with this at Autodesk 10+ years ago.

hi5 uses some pretty shitty tactics to achieve retention. Namely, fake users with "attractive" photos. Most profiles there are fake.

Rather than fixing the retention problem, they decided to hide it in the most ridiculous way possible.

The website also has an awful design.

hi5 (RIP) was originally a dating site and Pivoted to a Social Network.

Most profiles fake? I can say for sure that's not true. I maintained Postgres/Memcache/Graph DBs and the write load was real.

That said hi5 did engage in address book scraping and other dark patterns that you'd rather not see these days.

Fun fact: hi5 had a featured photos/profiles section based on popularity. Folks that ended up there deleted their account by 5x or more due to the unwanted attention their 'popular' photos garnered....

If that is the case then it was very poorly thought out. Am I interested in dating people that live in another hemisphere? no.

this was such an exciting time. there was such hope for distributed, secure social protocols. for interoperability, intertwingularity. we'd seen rss become powerful & loved, innovation like Wave (totally excellent off the wall great leaps & bounds), Buzz, & others seed the idea that more was possible, that we could expand how sites & users worked together across the web in a secure fashion.

OpenSocial was to be totally next level modular web. being able to build composite sites, sites built of tech from all over. we could enhance our experience on a site in ways that no site operator had pre-determined. It was going to be a totally new freedom for users.

this was really the climax of so much of the aughts, such hope vested in it.

however development felt like it went really really slow. time kept ticking. the API centric days of the web kept seeming to grow further off all a while, while this set of protocols felt under-developed under-supported; using the reference stack was something I tried a couple of times & never quite latched on to. i don't think alternative stacks ever came about, even though I feel like this was a protocol-centric approach.

a lot of this reflects my own likely deeply embellished memory. OpenSocial felt like it represented a lot of hope for a web that could emerge, a culmination of a lot of what was happening on the web, in fancier form, distilled. and it feels like the turning point where that- to me- far better web failed to really materialize, failed to become a thing. there were still a couple more years of good API based systems being the norm/expectation, but the wind was running out of the sails, each fief of the web collapsing back in to it's own. Google Plus launched 2011, not long after OpenSocial was underway, and until it's demise it never shipped an API for writing a post. you had to use their client to do almost anything. this was a ghastly shocking turn, back then, something that was almost unbelievable. now a days, clients like Signal &al, this is de-jure, the norm. systems are only what they were designed for you the user to get.

i think of opensocial semi-regularly. at the time it felt like the first big ride out for the "open web", like the beginnings of a new frontier. it is with enormous & sadness & disappointment that i report, here, a decade latter, that it feels like the last & final ride out of the spirited, hopeful, connective open web, that things only ever got worse after.

one last little side note, one of the main people i associate with OpenSocial, Dave Recordon, was just named Biden White House's Director of Technology: https://yro.slashdot.org/story/21/01/06/2129251/open-source-...

Oh man, Wave brings back memories. I’m still a little sad it wasn’t given more of a shot —- the idea of an open ecosystem of widgets sharing state using an OT (or similar?) backend is still something we haven’t seen done well, but in theory it could allow some pretty neat things.

It just didn’t scale well, at the time. I wonder how it would do on a modern browser —- or whether the backend was the bottleneck.

when I hear spec and xml I don't want to use that thing.

thank you xml enthusiast I hope you write xml all day long

Reply on news.ycombinator.com