How much decentralisation is too much?
shkspr.mobi
shkspr.mobi
One thing I really would like to have and get working, is a little selfhosted homeserver, instead of putting my websites on someone else's servers (although cheap). It becomes more peer-to-peer this way, and I'm only relying on an internet service provider. The best thing is already suggested by the article: how cool would it be if your smartphone (or pocket computer) functions as a server!?
Completely not cool, that is, pretty hot. And short-lived.
Mobile devices are only practical because they don't work 80-90% of time. Being awake for longer requires larger batteries. Being suddenly popular and thus withstanding much more load would just physically kill a node, by depleting its battery quickly and suddenly.
The luxury of peer to peer networks requires either a large power supply, or a limited availability of each node. If you introduce super-nodes with good power and constant connectivity, where weaker nodes can post their data to be cached and replicated, you just invent servers again.
> (not "an extra/old").
So then you concede completely, that anyone talking about these devices, would indeed be correct to say that it is cool to turn these devices into servers.
Great, you agree that there is a usecases for this.
FidoNet worked pretty well, as long as you did not mind your message to reach a destination within 3-4 hops by tomorrow. Maybe it's the right pace of communication in certain cases. Not tweet-style, but blog-style.
Super-nodes with around-the-clock connectivity also existed, advertised themselves as such, and were a great help.
There must be some scheme of symbolic addressing. DNS, of all things, is reasonably decentralized, it only requires SOA records for top-level domains to be propagated in concert. (It has other shortcomings though.)
I pick on DNS because it's an easy example of another thing you have to manage hosting your own content. Also, domain names are a hot mess and people mostly prefer to have identities ala Facebook, which could easily built into a decentralized protocol.
No you don’t. Not if the super nodes are almost automatic. The problem with servers is having to manually admin them.
What if you remove the battery altogether and use the phone as an alternative to a Raspberry Pi, instead of creating more e-waste?
If Android (or any other mobile OS) wasn't so locked down, we'd definitely be able to take more advantage of devices like that and not just for hosting stuff - they already have mobile data support, Wi-Fi, Bluetooth, a microphone, camera, GPS, accelerometer and so on.
In regards to load, my blog did fine with one of the articles on the front page of HN way back, and it was served from a container that has about 1 CPU core given to it and under 1 GB of RAM (albeit it's still a rented VPS otherwise).
My homelab server CPUs have a TDP of 35W and I use those for CI as well with no issues. My netbook has a CPU with a TDP that's around 15W and it can serve web apps with no issues, alongside Visual Studio Code and a desktop environment.
A phone is almost certainly good enough for simple workloads as well, like serving static files or interacting with a SQLite database, assuming that your content wouldn't get too popular.
But having it be a server while in your pocket is probably too niche of a use case, rather than having it on a shelf somewhere, connected to a power cable.
https://news.ycombinator.com/item?id=31841051
"I replaced Google Analytics with a web server running on my phone [2020]", 0 comments
https://news.ycombinator.com/item?id=28306681
No real point on my part here. There's just a lot of colorful discussion here and in sibling threads. I found these to be fascinating reads and they feel relevant. I believe the author was running the server from second link on their daily driver.
Author writes about using Termux as a webserver. Do note Termux's having to target older API versions if anyone is really contemplating the feasibility today.
(Sorry, mobile)
This is not a snarky comment. The thing in your pocket can, today, be convinced to act as a webserver and host your website. It just comes with all the downsides of having to host, administer, manage, and secure a website.
In practice, you can either have a secured computer or one that offers useful services over a network. Through decades of sincere effort and many billions of dollars of investment, it has turned out that you cannot make hosting, administering, managing and securing easy as pie and still have something that can do all the things users would like it to do as a server. We can't even reliably do this for non-server computers.
It's worth considering that sometimes our desires are in tension with another in ways that are not readily resolved.
No, in that you're completely correct that network positioning and exposure does not inherently increase the vulnerability a computer has. It does not introduce new flaws.
Yes, in that it's not just about vulnerability. A computer turned off may be in theory vulnerable, but in practice not exposed. Exploits happen when vulnerabilities meet opportunities. The more a computer is exposed, the greater the opportunity for exploitation. Making something publicly available, such as it offering up a website, is a pretty dramatic way to increase its exposure. As you might imagine, trying to perform useful work with data coming in over a network provides many more chances to try to exploit something than silently dropping all inbound connections. In extreme cases, you get things like the numerous worms of the 2000s.
It is precisely because of this excellent point you have made that modern networks are designed with layers of security. That there are other vectors is a concern you wisely and rightfully point to, but each vector is concerning and the ease with which any given one can be guarded or exploited is considered.
tl;dr: There's a substantial and very meaningful difference.
Sure; what matters is, roughly, the extent to which the computer is doing something complex based on data sent by an attacker. The thing is that these days people's pocket computers do so much complex work with user-submitted data, compared to low-level protocol parsing which mostly hasn't changed for decades, that I struggle to believe adding the latter would actually increase attack surface that much. Back in the '90s, where the main way to send data from computer A to computer B was for computer A to open a socket connection to computer B, that thinking made sense, but today almost everyone is consuming a firehose of data from random strangers via, well, almost every app on their phone and every website they visit.
My problem is mostly that the thing in my pocket runs Android or iOS, but I'm working on that problem.
> How well-equipped do you think your average person with a smartphone (or pocket computer) ...
Average people were pretty handy with computers and internet, even before the smartphone entered the scene and made it even more handy for even more people. So that would be the challenge in bringing some 'personal mobile server' to life.
I don't think turning them into servers is likely to improve this situation much.
As the cliche goes: "If I can do it, you can too" . If you have multiple sites I have no doubt you could get one up and running without much trouble.
I have a few RPI 3b+ and 1x RPi4B that I use and host from home. Mainly just to dink around with, nothing crazy.
One thing I did was get a cheap VPS ($4/m) to reverse proxy to my home.
My next iteration is I think I'm going to actually host the site on VPS but have my home network push static files to the VPS. My site is mainly for me and this way I can fully disconnect between home and VPS.
We will see, but again, this is why I have what I have so that I can play around and try different things out.
Best of luck to you!
> going to actually host the site on VPS but have my home network push static files to the VPS. My site is mainly for me and this way I can fully disconnect between home and VPS.
Smart. Thanks for this tip.
I've been looking into dynamic DNS services a bit, to see if I can/should use that for pointing to a webserver at home. Also, I'm considering to make a distinction between internetconnected but personal/private apps on one hand, and publicly served apps/sites on the other. Maybe I should just start and try to get a website online from home.
That will still expose your IP address, at least NOIP did when I was using it. NOIP was a good service but after 5 years I noticed that my IP never changed (yeah! took me a while to catch on! ;>).
The main driver for me was not showing my home IP on lookup and the yearly cost between a cheap VPS and NOIP subscription was roughly the same ~$60/year so nothing was lost there.
How exactly is that going to serve data when I'm in the subway or otherwise going through tunnels, or anywhere at all with bad reception...? Or when my battery dies?
Servers generally need to stay connected.
Unfortunately, the standard IPFS daemon tends to be a resource hog (otherwise I'd be running it on my Pinephone ;) )
I think the point being made here is less about new things and more that you should refine your requirements. The computer in your pocket can do anything a stationary computer can do - it's Turing-complete, so this should be obvious - subject to the limits imposed by the hardware. What limits are you, personally, willing to accept for your home-server-pocket-computer?
Good question. Food for thought!
I have been running servers on a "smartphone" for years, for example SSH servers, FTP servers, HTTP servers and authoritative DNS servers, with no noticeable heat or battery issues.
From a broader perspective of structuring our society so that censorship is technically difficult, we don't want to rely on a single provider. In that sense, we want both
* ISP should allow hosting from home that is fast.
* an ecosystem of website hosting providers, so ISPs are not a single point of censorship.
Is Twitter really over, though?
I wish people wouldn't try to make Twitter over by self-prophecy.
It feels disingenuous. If it were over, where are the stats?
Twitter seemed to be in decline 3 years ago [1], but its revenue seems to have been somewhat stable since 2020 [2].
Maybe declaring Twitter dead will never go out of fashion, and we just come up with different reasons.
Maybe "Twitter is dead." is the kind of statement that ultimately keeps Twitter alive, because you need platforms like Twitter to keep fake news alive.
[1]: https://www.forbes.com/sites/kalevleetaru/2019/04/23/a-fadin... [2]: https://www.businessofapps.com/data/twitter-statistics/
> Firing people with domain knowledge
From the sounds of it, many deserved to be fired:
> teams of Twitter employees build blacklists, prevent disfavored tweets from trending, and actively limit the visibility of entire accounts or even trending topics—all in secret, without informing users.
https://twitter.com/bariweiss/status/1601008766861815808?s=2...
Twitter was already, by all means, up in the air.
Twitter is a highly politicized environment where you get treated differently based on your views. Since it got a new captain, all those who were in power (meaning they didn't, on average, get treated unfavorably by the platform) are now claiming that the platform is dead.
I dare to say the erratic emotion is on the side of those who say Twitter is dead.
To be fair though, I really haven't seen a change in the content on Twitter at all since Elon took over (modulo meta-discussion about Twitter itself). Time will tell I suppose
What about the grocery store? Did you know that bad people use those as well?
Even going back to your bar example. I am sure that bad people go to bars all the time. And generally speaking, nobody gets kicked out, nor do people just leave.
You have your drink like everyone else.
So the way I read that opening is a tongue in cheek expression in the voice of someone who is a little overeager to promote the fediverse as the Twitter Killer (when it's just... a different thing).
Edit: Shared this thought only to say that there possibly is a different interpretation to this opening. Not to invalidate what you said (should have mentioned this in the start) :)
It really sucks. This is not what I wanted for the fedi when I started back in 2017.
It has lead to some admins forming "federations", silos, inside the fediverse.
Like I can see technical solutions for some problems, like DNSSEC verification similar to that of E-mail.
But there is no technical solution to snowflakes acting butthurt.
Of course, that didn't work when people started doing that whole "block X or we'll block you" guilt-by-association thing. But I just hope that's a minority.
I believe a solution to this may be more decentralisation, so a small number of admins with old and bitter feuds aren't able to control the experience for the majority of users on large servers. Admins should be friends, not celebrities.
That's why I'm not part of the whole UFoI thing, but I'm sympathetic to its existence.
As long as people follow the server rules, aggressively presented arguments, anger, fear, rude words said to famous accounts, should not lead to de-federation. Our rules already disallow all forms of xenophobia, transphobia, attacks, doxing and all that stuff.
End users need to take more responsibility for moderation in the fedi. That's why we have the Mute button. But people would rather take "action" with the Report button, than the Mute button.
Because they get so angry at some stranger online. You don't know anything about that person, or what they're going through.
I have experience meeting reformed nazis, previously violent people who are trying to reform. So people go through stages in life, and maybe sometimes they're a bit off, but some day that person is going to wake up and then we should be able to help them out of their daze. De-federation makes this impossible.
You can't have a technical solution to that, it's a design choice/tradeoff. You can have a highly permissionless social network, but then it's not useful for the average person because it doesn't exclude the "undesirable" content; and you can have a highly regulated social network but then it's obviously not useful for people who do want to make the excluded content; but you can't have a single social network (even if federated) for everyone, splits are inevitable.
If such a feature was more robustly built into a platform like Twitter then the company themselves would have a lot less pressure in making a universal decision (which guarantees some disagree with)
'snowflakes acting butthurt'
('_')
I'm not Apple defender at all, but if they did that I would say chewing battery unnecessarily is a pretty good reason. Of course they would likely have other reasons, actually not good ones, too.
The state police doesn't really have oversight over the local departments. In fact, I don't think anyone really has oversight of local police departments except possibly for local voters.
They are all bound by the various laws, but within those laws they have tremendous amounts of flexibility.
It's in their self-interest to cooperate, but they don't have to cooperate if they don't want to.
> In fact, I don't think anyone really has oversight of local police departments except possibly for local voters.
this does not feel right at all, especially with all I read about the problems with the police and the lack of accountability of the actions of individual police officers. I do not feel confident in directly overseeing something as complicated as a police force. What do you do if you feel the handling of your police force is inadequate?
I live in a small town, with a handful of police officers, and their boss (the commissioner) is elected by me and my neighbors. There's no greater accountability than that!
People hate POTS. You're identified by a number. It includes a (now meaningless) area code. Portability is a problem. You can switch carriers (although prepaid portability can still be an issue) but it's a process. Also, whatever portability you have disappears when you cross a border.
POTS is filled with spam. Just like email (another federated service). As soon as something is federated it becomes incredibly difficult to police bad actors. If you, as a telephone exchange has too much spam (eg robocall) traffic, you might find yourself blacklisted by large players like Verizon and AT&T. So what do you do? You "launder" such traffic with enough legitimate traffic so as to not trigger any enforcement while charging extra for known offenders. Because that is profit.
A federated system is inherently a game of whack-a-mole with bad actors while being a race to the bottom.
Messaging services like Twitter, FB (Messenger) and WhatsApp all grew and continue to exist because of centralization while federated alternatives inevitably always die or simply never gain real traction. There are real efficiency issues. For example, when policing offensive material (eg CSAM) is each entity responsible? How is that efficient? They might have to do because given material may never leave a given node.
The question you should always ask is: what problem does this solve for the user that they care about?
Related: https://berk.es/2022/11/08/fediverse-inefficiencies/
Yes, some of them were actual trolls or spammers, but many were just ideologically misaligned with Twitter the company (and/or their advertisers).
I think the problem to solve is how can we allow people to communicate AND DISCOVER other people's opinions without a particular company/organization making that decision for them.
Websites aren't good enough because they aren't very discoverable and don't have a robust discussion mechanism. I don't even think Mastodon allows this because admins seem pretty opinionated and end up cutting out chunks of the fediverse without the users' consent.
I'm not sure what the solution is but that is a real problem and it does seem necessary to "decentralize" to some extent as every group will always have conflicting opinions with other groups
The thing is that most centralised platforms has been really bad at incorporating cooperation so I can understand if some people are leaning on decentralization being the only way to reach this.
Like others have done in technology, they can become initially equal members, then degrade the others, and finally make it painful to use/participate with other members of the federation. This would force federation members and users to succumb and join the empire as a subject versus equal.
For now, I'd say that was probably because it's a new thing to most people and people hear "Mastodon" and, understandably, head for mastodon.social. Give it a year and I think people will be much more spread out than they are now.
Gab (the Nazi thing) and Truth Social (the Trump thing) are both based on Mastodon, but nobody wants to federate with them.
0. a twitter clone backed by a few billion VC dollars will emerge with the promise to continue the practices described in the twitter files, and the refugees will flock to the warm and familiar embrace
Thsts entirely anecdotal of course, but I think phrasing it as an issue of the newer generstions doesnt encapsulate the whole problemspace.
If anything id say those who didnt grow up with tech is even more susceptable to dark patterns.
Do we need electricity? ...
Do we need writing? (Plato, 600BC)
(btw, anigbrowl, in case you haven't already seen: https://news.ycombinator.com/item?id=33942280 )
We've got Mastodon, and it's doing a bit better. We had Hive, which did okay til the security issues and shut down. Cohost and Post kinda... exist I suppose? And there are a few more I can't remember.
I've got an account on each one, and about 1/10th of those I follow exist across all of them combined. Maybe most people just don't care about a Twitter replacement and are fine without one.
Firstly, a significant fraction of the most interesting accounts I follow on Twitter have moved to Mastodon.
Secondly, Mastodon feels incredibly fresher than Twitter, which has completely ossified.
Thirdly, I know this sounds slightly contradictory, but it really is quite a lot like Twitter, in terms of recreating the good aspects of Twitter.
* Advertisers leave to protect their brands
* Major multi day outages
* Data corruption
* Security breaches
* Outlandish payment schemes for posts or reach
* Company is broken up and sold off in chunks to pay down the financing debt
But I do question whether people will move towards any particular alternative, or whether they might even stick with other services they use instead. None of these replacements seem to quite have the pull necessary to attract most of the existing Twitter userbase, and none seem to be taking over in the same way Reddit beat Digg or Facebook beat Myspace. Perhaps the market for a Twitter esque microblogging service might decline as people spend more time on Discord/Reddit/YouTube/TikTok/whatever instead?
Hmm, I guess the only way to know for sure is to wait and see what happens when Twitter comes crashing down.
Mr. Musk also continues to cut staff and leaders, including Nelson Abramson, Twitter’s global head of infrastructure, and Alan Rosa, the global information technology head and vice president of information security, according to four people familiar with the moves.
What's the optimal size of a server?"
The facts indicate that we just don't know and it's likely that we will not know for quite some time until we've a much better handle on and understanding of the long-term statistics about how connectivity and centralization impinges upon human behavior and how humans socially adapt—or do not do so.
That's to say we need to know both the positive and negative impacts on society of the internet with reasonable accuracy before we can make any authoritative assessment and I reckon we are nowhere near that point yet.
Let's look back a little. Until the 1970s the connections between people were reasonably stable and established, radio TV and newspapers were usually conservative and reasonably accurate in reporting - sensationalism in reporting wasn't yet commonplace. Then in the 1970s we had the arrival of radio shock-jocks, and beat-ups and sensationalism in television news and newspaper reporting became commonplace. This in-your-face type of media presentation altered the way the population thought and behaved, and the fact is we still haven't fully analyzed or accurately know the effects that those changes brought to society over the past 50 years.
The internet and social media have brought about social and political changes to the way people think and act and similarly to the collective social fabric and those effects have been much faster and dramatically greater than those that occurred in the immediate pre-internet era.
This is a huge subject and we've only just recognized how important this issue is. Next stage is to gather even more data and it'll be years before we have any definitive answers.
And we can't take for granted that society will evolve towards continued progress because it clearly won't without advocates.
Remember back in the day, when you just downloaded limewire or kazaa and that was all that was needed to get on a decentralized file sharing network.
Compare this to the recent experience of many people deciding on a mastodon server and joining it and figuring out how to follow people from other servers.
Such a thing would make joining and using federated systems much easier in my opinion, and the easier it is, the easier it is to gain adoption, and the easier the adoption, the more likely to gain traction.
When I chose a social media platform in the past, I didn’t care about engagement, or who was there or how to find people, or the philosophy behind how it operated, all I really cared about that the content. I didn’t even care about the content being moderated. Why would I care about Nazis using Tumblr if I didn’t follow any?
I think that some people have become almost dependent on the concept of online social networks. Yes they are cool and all and its cool to find ideas and conversations online, but I think we are heading down a road where that intangible taste and intrigue is being lost because certain factors are being prioritized to the point where people are forgetting that if they just cared a little bit less, they could disregard the whole jumble of protocols and policies and start a message board.
But other people, possibly young and impressionable people, will.
The thing is that, if I were to affect my beliefs and moral coordinates as an expectation on social media and the channels of public discourse…I would have to settle on just a message board. I’m using message boards as just an example of a possible solution where the desired control is there minus federation (which I don’t care about).
If we all were to affect our beliefs and moral coordinates as expectations onto social media and online discourse…what would come about?
This is a conflicting topic to me. I’m aware that I’m speaking into the void here now, zimpenfish. I don’t expect you to respond or read where I’m going with all of this but you’ve got me going now.
All in all I just think that there are some issues present online that will not be solved online, necessarily. On one side are the real life matters relevant to society and our well-being. And then there are things like “open standards”…and then there’s everything in between that connects the two.
I think the message board idea is a fine one.
This guy clearly hasn’t heard of the nazi-at-the-bar story. (Paraphrase: an usassuming guy wearing a white supremacist symbol sits and is immediately told to get out, with violence threatened if necessary, by the bartender. The bartender later explains there’s a strategy for unassuming authoritarians who don’t cause a problem end up bringing their friends and suddenly they’re driving other clientele away because it’s a nazi bar, and then it’s too late to peacefully resolve the problem because the nazis will use their numbers to fight any enforcement.)
Setup and maintenance of the mailcow dockerized stack is quick and easy. Add a $12 VPS of your choice to it, et voila.
I recently swapped email to protonmail from gmail. Why? Because I read an article about the Terraria developer just getting banned and losing his email for no reason. “That could happen to me!”
After a long while of using proton (9 months or so) I got really tired of the lack of ecosystem (at this time there were no public apps on iOS or android besides mail) and I still had to use google for my contacts sync, notes, etc.
So what I ended up doing is moving all my shit BACK to google from proton (massive headache by the way) and then just self hosted my own email address domain. That way if google ever decides to ban me, I can hop over to Cloudflare and point the email domain to another email provider and violà!
One of my main complains with protonmail was that the search function actually kinda sucks. GMails search is so good, and the filters you can make are so easy, I could go on. That’s why google is the industry leader! But why does proton make me feel more comfortable? Zero-trust. Proton doesn’t know what is in my emails at rest, but google does.
So I propose the idea: why decentralize? Compare mastodon and Twitter (pre-Musk). The whole reason that Twitter got huge is that everyone is on it. I loved the old school “everyone had their own message board community” internet, and IRC was really cool. But the VAST majority of users don’t want that back. It’s much more convenient to load up discord to talk to people than an irc client that has like 40 different segregated communities, in the same way that it is much easier and more convenient to use Twitter than to manage all of what mastodon is.
Even then, look at something like Session (the username only fork of signal). I’m not 100% on the details, but they claim to be decentralized, but yet I don’t have to input server details. That means that there has to be some sort of centralized database or application or table that tells me where all of the servers I can connect to are (or perhaps a first server that you connect to that then spreads to another server etc)
You have 4 choices in the modern world:
google (nothing protected, totally centralized, but the most feature rich and fastest)
protonmail (centralized, protected, but not as feature rich and fast)
session (decentralized but there’s still a layer of centralization to make it easy to use, but still more complicated than the first two options)
Mastodon (fully decentralized but wildly inconvenient for the average person, significantly more difficult to use, and likely will never adopt the number of users that Twitter has)
Remember, we are the minority. Google is the best for most people, as is Twitter, as is Discord, as is iMessage or SMS, etc.
People use what is the easiest and most feature rich. Decentralized stuff will never meet all categories of feature rich, fast, and easy to use. It just isn’t possible, and if it was it would be the biggest smash hit in history and we wouldn’t be having this discussion.
That aside, all email is pretty much as decentralized as Mastodon is, since it's all federated.