Hacker News Reader

Comment by chc4 | Hacker News Reader

This sounds like what I wanted from GNU Social when I first joined over a year ago. GNU Social/Mastodon is a fun idea, but it falls apart when you realise that you still don't own your content and it's functionally impossible to switch nodes like it advertised, along with federation being a giant mess.

I tried to switch what server my account was on halfway through my GNU Social life, and you just can't; all your followers are on the old server, all your tweets, and there is no way to say "I'm still the same person". I didnt realise I wanted cryptographic identity and accounts until I tried to actually use the alternative.

That's also part of the interest I have in something like Urbit, which has an identity system centered on public keys forming a web of trust, which also lets you have a reputation system and ban spammers which you can't do easily with a pure DHT.

View on HN

Not being able to switch nodes pushes you to try and host your own instead. That's what I've done. IMO we should instead be looking at packaging a self-hosted version into a native Windows and Mac app. Run it in the background and everything's done.

This is what I want. I've been wanting to build something like this for a long time. Something where I own my data. I can back it up and if my laptop gets stolen, I just import my data on a new machine and we're good to go.

The challenges that I see: - Making it easy for any user to get up and running. - De-authenticating old devices. - Making it available from any mobile device.

I bought a VPS from CloudAtCost for a one-time fee of $35. Set it up with nginx and GNU Social and pretty much haven't looked back since. My instance is https://kwat.chat.

Ideally, the end solution would be dead simple. Download the Windows app, run it, put your credit card in if you need a URL registered, and it does everything, including daily backups to a folder on your disk.

That's basically the aim of Urbit. An instance you control that you can connect to other services to tie those external identities together.

Sounds a lot like Diaspora as well.

And then your laptop gets stolen and everything's gone.

And then you remember you set up an automated encrypted backup to the cloud and thank your past-self. Now is the time to do it if you're not doing this:

https://msol.io/blog/tech/dirt-cheap-client-encrypted-online...

Why, oh why do we have smart individuals replacing internet with the word for the visual analogy we used to represent the internet to not knowing better people in position of power ?

This is not a damn cloud ! This a remote computer you can access over the internet. Can we stop with this use of marketing lingo please ?

Sorry buddy, that ship has already sailed. Terms become popular because they are useful. We all know cloud means "a remote computer accessed over the internet", but that is rather cumbersome phrase. "Cloud" says it in 5 characters. Can you suggest a better term?

As a fan of both linguistic games and playing devil's advocate, how about:

"server"

6 characters, sorry ;-)

"host"... 4 characters, beats cloud...?

"Cloud" refers to a commoditized service, generally highly available / performant.

"Host" is generally less ambiguous, referring to a specific thing given the context of the discussion. It's a pronoun for machines (kinda).

Terms become popular because they're useful, but useful to whom?

The purpose of the original coinage of the word "cloud" was to obfuscate that you really meant "someone else's computer". It gives a nice warm, fuzzy decentralised impression - clouds are natural and ubiquitous! No one owns them! If it's in "the cloud" (note the definite article) then it's safe in the very fabric of the network, right?

Nope. It's in Larry and Sergey's basement. Not decentralised at all. Just somewhere else.

The proper term is "server", "datacenter", or "network", depending on what you're actually trying not to say.

If I understand scuttlebut correctly, your stuff is broadcasted to whoever it might concern and the pubs. If you still have your private key, you should still be able to access whatever is in the ether, right ? You somehow become the recipient of your own messages. Only the thief will also have access to your private key, so the account can be considered compromised.

This is a valid point and I don't trust my computer. I would trust, however, a Ledger Wallet http://ledgerwallet.com/ and theoretically and economically it's feasible to have a Ledger wallet app to sign every SSB message. This would be awesome to have.

so, data breach means all your private data is irrevocably publicized.

What percentage of users do you think would be affected by such cases? If it's something over 0.001%, it's a huge problem for a social network.

Sites like Coinbase and Github exist because they re-centralize distributed systems — users don't trust themselves to host their own data securely.

Alternately, if this isn't a problem, why don't users simply host all their own infrastructure for existing tech problems today ?

I'm sure someone capable of living in the Mojave Desert is capable of hosting their own infrastructure - is this network simply for those people, or is it also for journalists, trans people, and HR professionals?

Yep, that's how it works.

Make backups. How is this any different than saving photos just in case you spill water on your laptop keyboard?

99% of users either don't have backups of their precious photos, or only do because they blindly clicked through "set up iCloud" or their Google/Android or Windows equivalents.

It's different precisely because most social/cloud networks abstract away backups.

Very few people have considered whether or not they should attempt to back up their Facebook account. Same's true for Flickr, Twitter, and Gmail.

Frankly I trust Google and Facebook more than I do myself with regard to backups. I know it will eventually burn me, but I've lost, misplaced or misplaced the key to my backups more than once.

I'm probably in the minority being so irresponsible with my own backups, but I'm not alone.

Google and Facebook have a lot on the line with regard to user trust of their reliability. Also, they can't monetize data that they've lost.

But goosebook and facegle make backups for their own sake. You're still tied by vendor lock-in and can be locked out of your own data on a whim and prevented from switching to another service provider. They can and do monetize data that users have been locked out of.

I'd rather have my own backup copieS and take my own responsibilities. The scenario you evoke here would not happen if you had proper backup strategy, two is one and one is none.

The 'cloud backup' part could still be extracted away, using integrations with common providers (dropbox, onedrive, gdrive, etc). It would be a configuration step, but one that has pretty obvious benefits to the user so maybe they'd be likely to supply their credentials for that.

Maybe they should?

Maybe but the person's point stands. Facebook has multiple datacenters with probably some kind of backups. It keeps things for years at a time even when it doesn't need to. It likes to because it helps the business model. Hardly anyone's pics and stuff will disappear.

Compare that to their experience at home with personal gear. Many like the convenience and reliability of Facebook over their own technical skills or efforts. You'd have to convince those people... a shitload of people... that they should start handling IT on their own. Also note that there's many good, smart, interesting, and so on people that simply don't do tech. Anyone filtering non-technical or procrastinating people in a service will be throwing out lots of folks whose company they might otherwise enjoy.

So, these kinds of issues are worth exploring when trying to build a better social network.

It only takes one time of having your account locked by facebook or google and said people are automatically convinced of the obvious advantage of being in control of your own data.

Same with backups, lose your data to drive failure or theft once and suddenly having a backup strategy becomes a priority.

But as long as they have not been bitten once they don't care enough to actually do something proactive.

They usually resolve those with access to their data. Their computers getting trashed by malware or breaking is different. It can cost money to do recovery that might give them nothing. That concern is the more common case.

My laptop keyboard has drain holes directing the liquid to the bottom without ever getting inside, so that's one way it is different.

Or you shut your PC down when you leave the house and then want to access your network from your phone.

SSB's central premise -- distributed users, making an ad-hoc network connection whenever they are physically close, or perhaps have some network connection -- bakes in an assumption that a user's ability to connect to the network is sporadic.

It seems like the system would work just as well for people who decide to turn their system off when they go to work, or are on a sailboat. Of course it's not convenient in the same way that always-on social networks are, but that seems to be specifically not the point of SSB.

How about this: you buy a physical device at Wal-Mart for $29.99, plug it in, hook it up to your wifi and leave it plugged into an outlet. It's got Mastodon or GNU Social on it and could look like this, but branded: http://thegadgetflow.com/wp-content/uploads/2015/10/SmartPlu...

And then my internet connection goes down. Power goes out. I run over my data cap for the month. Comcast shuts me down for running a home server. My home network gets DDoS'd. I miss a patch day and I get hacked.

None of these things are a concern on traditional social networks. They have to be solved before the world has any chance of moving to a decentralized network.

They have to be solved before the world has any chance of moving to a decentralized network.

I say no. Most of them don't have to be solved first.

Feel free to convince me that 3 days of downtime on my personal messaging account cannot have my personal account is a problem.

It doesn't matter how you feel about it, take a look at people complaining when Google put a news article about Facebook at the top of the results instead of the Facebook login page:

https://www.theguardian.com/technology/blog/2010/feb/11/face...

These are people who typed "Facebook Login" into a Google search, clicked the first result without reading, and got confused. Now tell these same users that Comcast blocked their social network or that they can't log in on their phone because their home Internet connection is down.

If you want a social network filled with just people like you and me, look at App.net or GNU Social for inspiration. If you want average users to sign in, these issues absolutely do have to be solved.

Still not convinced. It doesn't need to be for everyone in the beginning.

The early web (when I entered and before) wasn't for everyone. And thats OK for me. Actually I think it is a good way to start.

"he early web (when I entered and before) wasn't for everyone. And thats OK for me. Actually I think it is a good way to start."

It got where it went by doing the opposite of what you're suggesting. The walled garden for smart elites were mostly working on OSI from what old timers tell me. The TCP/IP, SMTP, etc involved lots of hackers trying to avoid doing too much work. Much like the users you prefer to filter. Then, it just went from there getting bigger and bigger due to low barrier of entry. Tons of economic benefits and business models followed. Now we're talking to each other on it.

The problems were solved with centralization.

For this tbing I don't care about those problems.

The early web was devoid of average users. Then the flow of newcomers and not knowing better users surpassed the old timers and knowledgeable users. Then we entered a leveling downward race to a web tailored to their needs because they're the large majority.

What do you call traditional social networks ? To me a traditional social network is an AFK thing.

If you're internet connection goes down, power goes out, you get DDoS'D this would hinder your ability to use any third party online service anyways. The data cap and ISP restrictive terms of service are a different problem that would be challenged and fixed given internet subscriber would go the p2p self host way. The commercial ISP situation is a terrible mess right now. If you got hacked unplug from network, boot from recovery, restore from backup, you're back online in less time than it takes to recover a hacked facebook account.

You say decentralized but it seems to me you meant distributed here.

It's not that hard for the hypothetical manufacturer to set `git pull` and `apt-get update && apt-get upgrade -y` to run on a cronjob every morning at 4am.

Putting another barrier in front of it is not what's needed for people to use it. Treat it like email (or heck, early Facebook), get big clusters of users in by convincing universities and colleges to run a campus server. Businesses would also be a good idea but a harder sell

I do quite like this idea. And, it has precedence.

Then I have yet another device permanently plugged in and running, at a time where I and frankly all of us should try to reduce our energy consumption.

I'd rather see a universal single consumer server with easy download and plugins for all of this stuff. Host my social network, my mail server, my cloud apps, etc. Basically, make social network a part of OwnCloud and sell OwnCloud boxes. Instead of a million small devices, I do one big one... and "big" can still be RaspPi.

A Raspberry Pi is very low-powered though, it's completely insignificant compared to the power used by the webservices you use daily.

If the power consumption of a RPi for each household with someone like us is a major thing then I say we have come pretty far in reducing waste of energy. :-)

People don't want another thing to plug in even if it's low-power. IoT is still nascent and I already don't have enough outlets in my house...

I disagree. Google uses 1/4GW worldwide and that's less than a watt per user.

Where does that number come from?

One data center * 24 backup generators * 3 MW each = 72 MW per data center. Four of those = 288 MW > 1/4GW.

Google has more than three data centers.

Ah yes, that number is from 2011, published by Google. Can't find the original. But it was widely reported[1].

Assuming that they're doubling energy consumption every year they'd have reached 8GW in 2016. That's 8W per user if we assume 1 billion users. Energy usage of a Raspberry is not insignificant relative to even this.

Doing things at scale is vastly more efficient. And only a subset of Google services can be relegated to a Raspberry. Even if you host your own mails, are you ready to ditch the Google search index and Youtube?

[1] http://www.nytimes.com/2011/09/09/technology/google-details-...

It's always going to be running somewhere.

Ha. Good joke.

> Not being able to switch nodes pushes you to try and host your own instead.

And you just use it for yourself? well, ok... but at that point you could also use a fully decent system.

You can follow and reply to any one on any node from your own system. Your feed is populated from people on any node you follow.

> [...] which also lets you have a reputation system and ban spammers which you can't do easily with a pure DHT.

Sounds interesting. How can you ban spammers when they can just create a new public key/identity if their old one is banned? And also, what does "banning" comprise, in a decentralized social network? I would assume it would be sufficient to just "unfollow" that particular identity.

It seems like you could get a some of the efficiency gains of having lots of people on one node, but avoid the difficulty of moving, by setting up a new node for each person even if they are hosted on the same box. That way you can pick up the whole node with your account and move it, instead of trying to move your account from one node to another.

The Fejoa project (https://fejoa.org) actually targets this problem. It aims to make it possible for users to change their hosting server without losing their contacts.

I mean, my view on the switching nodes thing is. It's not like you can just switch emails. Sure you can install a shim in there to forward everything, but there is no way to actually switch it. And that is the design GNU social uses.

Yeah, and that's also a pain point with email.

It's a pain point with the entire design of federated services. On the other hand, the pain point with the monolithic services design of Facebook is I can't even talk to people on other services so...

I'd rather be able to email everybody and have it be annoying to switch than be able to only email people on my chosen provider (and then have to make an account on every service anyway).

Reply on news.ycombinator.com