> Thus saying “People don’t want to run their own servers” is akin to saying “People don’t want to start their own YouTube channel”. Both sentences contain the same amount of statistical bullshit.
The key difference is that starting a Youtube channel is free and dead simple. The reason people don't want to run their own servers is because in most cases it's as expensive, or more expensive, than having someone else do it for you, and requires a huge amount of skill and patience, not to mention risk tolerance.
In general, I think this article misinterprets the statement that "People don’t want to run their own servers, and never will" by taking it too literally. The meaningful counter-argument isn't "this cannot be factually true, some people want to run their own servers", because that's clearly not what he meant. Obviously some people want to run their own servers, and some people have to run servers to keep the whole thing going, but that's a trivial point rather than a devastating counterargument.
> The reason people don't want to run their own servers is because in most cases it's as expensive, or more expensive, than having someone else do it for you, and requires a huge amount of skill and patience, not to mention risk tolerance.
It very much depends on what do you need a server for, and may not always be true.
However, IMHO primarily it's neither of these reasons you mention, nor any rational reason at all, but simply the fear of getting out of the comfort zone. The same reason why majority of people are not into DIY, but rather pay others to fix their plumbings or pour the concrete or change the lightbulb in a car. If you know nothing of it, running a server on your own sounds scary, you fear you'll screw up something, it's all stressful, and you'll rather pay to make it someone else's problem.
But then you also have a not-so-insignificant number of people who really enjoy in DIY approach and love doing as much as possible themselves. So the author's counterargument IMHO is more about Moxie stating something as an absolute truth, while in fact it's more like "majority of people will probably not want to run their own servers, under the current state of affairs". However I'm old enough to remember the time when the same applied for Internet - majority of people were not interested in messing with modem drivers and PAP/CHAP scripts just to connect to some BBS to chat with people, the idea seemed as ridiculous waste of time if you asked my father. And yet here we are now just some 25+ years later, whole world is connected. So perhaps the centralized platforms can't be avoided, or perhaps we need to make running your own server easier? And maybe both ways can co-exist, because different people want different things?
Napster worked perfectly fine without any config. Old Opera Unite showed a world where untrained users could start safe servers. Tools like Popcorntime show that such tech continues to work fine.
People aren't given the choice, because there is so much more money in centralized, paid services. They've completely crowded decentralized tools out.
> However I'm old enough to remember the time when the same applied for Internet - majority of people were not interested in messing with modem drivers and PAP/CHAP scripts just to connect to some BBS to chat with people, the idea seemed as ridiculous waste of time if you asked my father. And yet here we are now just some 25+ years later, whole world is connected.
There's a vast gulf between running a server on the public Internet and writing a modem script. When writing modem scripts was necessary to get online very few people actually bothered getting online even if they had a computer capable of doing so. Running a server of any type of the Internet is rarely a trivial exercise. Even when running one on a home system from a residential ISP you've got to poke holes in your router's firewall and maintain the server hardware and software. These are not things most people are even capable of doing let alone being interested in it.
> So perhaps the centralized platforms can't be avoided, or perhaps we need to make running your own server easier? And maybe both ways can co-exist, because different people want different things?
No matter how turn-key you make "running a server" there's going to be technical issues many people won't understand or be able to handle. We already live in a world where turn-key Internet connected devices are exploited en masse by malware.
You're also ignoring one of Moxie's major points that a huge portion of the population gets online via mobile devices. Just the basic design of mobile OSes largely precludes long-running server software running in the background on them. They're also on CGNAT networks with varying IPs and roaming on and off various WiFi networks. The only meaningful way for them to serve content is to bounce it off of centralized servers that don't have the same limitations of power or network connectivity.
Turnkey systems are a special case, because they specifically hide the configuration knobs that otherwise might be twiddled by even ordinary users (and thus keep the landscape from being a monoculture).
This idea that you have to hide everything a user isn't expected to need to understand until they prove they already understand it, which seems to have originated with the Mac, is stupid & gets in the way of gradual & natural mastery. Systems that present configurability while having reasonable default behavior invite users to explore them at their own pace, and inevitably lead to ostensibly "non-technical" users gaining whatever specific technical knowledge and skill benefits them directly. Even if they end up making a misconfig, a million random misconfigs is very different (in vulnerability terms) from a single unfixable configuration hole deployed to a million black boxes.
A question that people in the computing field need to consider: is computing more like plumbing or more like reading and writing? It's an important distinction for many reasons. I would pay someone to do most of my plumbing. I wouldn't even think of outsourcing most of my writing. Should computing always remain a scribal practice as it is now?
At the end of his review Moxie seems to indicate he feels this shouldn't be the case. I don't see how that aligns with re-defining what it means for "everyone to run their own servers"
I host my own server. It’s fun! I get to play with different software configurations and sharpen my skills.
But I would never want to do it for business. Or for anything large scale. I also never want to be my own bank and hold my wealth in a crypto wallet. I have a better infosec knowledge background than 99.9+% of the general population, which is how I know with full certainty that I don’t want to store my wealth in a way that would be instantly destroyed if I lost the keys or they were compromised.
I can guarantee you that number is less than a million people. And considering internet hosts Billions of people, that is less than .1% of people in the world. We can safely generalize people don’t want to run their servers
And how many of those billions of people are using Internet in read-only mode, not creating any content? Don't know the percentages, but clearly it's a big majority, and that majority doesn't really need any servers on their own, there's no enough incentive for them in it.
But as views on the importance of privacy change, perhaps that will change too? Perhaps people will want an extra layer of protection, if it can be made easily available and convenient enough to them? Any modern phone's hardware right now could easily run a number of service nodes in some distributed network, if there was such a thing and if their use could be made fairly transparent to ordinary users.
Right, but then what’s the selling point of a “decentralized internet” if like you said, most people have no incentive to have servers of their own. Decentralized internet is what is being sold in web3.
Improved privacy and better control over owning your own content. Both of these things, of course, are currently looking completely irrelevant to like 99.9% of people online, but as the surveillance increases exponentially this might change. Or not, we'll see.
> And how many of those billions of people are using Internet in read-only mode, not creating any content? Don't know the percentages, but clearly it's a big majority
I wouldn’t be so sure. Each day there are 4.5 billion items shared on FB (including 350 millions new pictures) and 500 millions new tweets.
> in most cases it's as expensive, or more expensive, than having someone else do it for you, and requires a huge amount of skill
What is expensive? Hosting from home has a lot of benefits if you’re not behind CGNat and how we used to do things when I was 13. If a 13 year old can figure out port forwarding and DNS I’m fairly certain you can.
I’m aware there are drawbacks but a $5 VPS is not outside the realm of reason if you want an additional level of reliability and are scared people will ddos your line.
Sysadmin skills are so easy these days they’re forcing developers to do it as an additional part of their responsibilities.
Either it’s easy: and everyone should do it.
Or it’s not: and we should start bringing back sysadmins.
> If a 13 year old can figure out port forwarding and DNS I’m fairly certain you can.
I think the truth is the reverse. To me, it feels much more likely that a curious 13-year-old with plenty of time on their hands can figure out port forwarding, than a middle-aged (or older) adult who has little technical skill and little desire to develop technical skill. And that probably encompasses most of the people on the planet.
> I’m aware there are drawbacks but a $5 VPS is not outside the realm of reason
$5/mo is absolutely too much for many, many people on Earth.
> Sysadmin skills are so easy these days
I feel like you are incredibly out of touch with the average internet user.
You are missing the point, it is unreasonable to expect everyone who wants to participate to be able to run their own server. It is too specific a skill.
Imagine if you had to learn how to be a mechanic in order to own a car, the number of people who would able to dedicate the time to own a car would be far fewer.
My mom wouldn’t know what the hell an NTF even is, and if I tried to explain it to her she surely wouldn’t see any value in them and wonder why anyone would be stupid enough to pay for one.
The big expense is time and the third way is that we scale up by having sysadmins and developers create services that are significantly easier to use and maintain as a whole.
If you're looking for a solid example of this look at Instagram which at the time of it's sale had 13 employees and ~10 million users.
Even if all 10 million of those folks had the expertise to run their own server, keep it updated and secure, keep backups, etc. it's still just wildly more efficient to have it centralized and hosted for you.
But we're not talking about people running big websites, we talk about people running stuff like pi-hole and mastodon instances. It's all very rough around the edges right now, but the idea of distributed platforms is not to eliminate big sites run by big companies, that will always exist, but we could also have people running micro services for their own needs right on the devices they already use (phones, tables, laptops, nfs boxes), without needing any extra expertise for that, just like they run a phone app right now to check email or weather.
It's rough around the edges for someone who is technically skilled and has some experience in system administration.
For others (the vast majority of internet users), it's a nigh impossible task.
I feel like everything in this space is perpetually "rough around the edges", and has been that way for decades. I don't have confidence that it'll be any different this time.
> Sysadmin skills are so easy these days they’re forcing developers to do it as an additional part of their responsibilities.
Nah most devops is IaC and sysops with very little if any dev. As some one that edits yaml files all day, I will die on the yaml is not "dev work" hill.
"If a 13 year old can figure out port forwarding and DNS I’m fairly certain you can."
Go ahead and take a poll of people who would want to do that, even after understanding it's 'relatively easy'.
I'm in tech and the thought of having to maintain my own infra makes me recoil.
You can build your own house, fix your own car, do your own plumbing, run your own servers if you want, but if you do, it's most likely for some niche reason, not because it's convenient and most people probably just don't want to go with this approach.
It's a very different kind of skill and risk tolerance.
If your YouTube channel doesn't take off, you've only lost the time invested.
If you misconfigure your server and it loses data or is hacked, it might seriously disrupt your personal life. (Imagine someone taking over your email server and stealing your identity.)
In both these cases, there's an entry level that people can get into without much skill, effort, or risk. (Running an email server, of course, is not part of the entry level: email is heavily gatekept.)
Most people who upload videos to youtube are going to have less than ten views total on those videos, which is fine, because why should they maximize views on a crappy phone video of their cat or something? Likewise, you can run lighttpd with little to no configuration in order to make some directory available, & nobody's likely to attack it even if it happens to be vulnerable, simply because most people are not hosting anything valuable enough to be worth exploiting.
The ideas we have of running a youtube channel or running a server are both affected by the hypervisible minority of highly polished professional versions: when we think of youtube videos, we're more likely to think of Hank Green than x_greenfan77_x, and when we think of web services, we're more likely to think of Facebook than we are to think of that python script we dashed off in ten minutes that's been working unobtrusively for eight years. The difference is that running a server has a mostly unearned reputation for a higher minimum technical skill involved: everybody knows that you can record a youtube video with no makeup & shitty lighting and sound, because everybody has seen amateur youtube videos, but not everybody knows that their torrent client is a server.
You are missing the point completely. Someone who knows a lot about makeup and spends lots of skill and patience but in a completely different domain. Yea Linus tech tips runs their own server because they happen to be in that domain. But that's not enough people to create a thriving ecosystem.
I think I'm in a unique position here to reply to this.
I run my own servers, and my wife is a professional makeup artist.
The amount of time, energy, and effort she puts into just ONE single Instagram post is more than I put into any of my servers.
It took both of us years of learning to do both these things we do, and honestly I get that you picked makeup as some kind of example because you think that makeup is easy or something.
But it's not.
Instagram, or YouTube takes a lot of skill, skills in recording video, camera, marketing, lighting, these aren't makeup skills, these are skills on top of makeup..
Then there's the outlay for recording and lighting equipment. Tripods to hold the camera, microphones for good audio.
If you don't have these things your video will look like crap and people will click off it.
Then there's the thumbnail! That's a whole thing on its own.
YouTube or Instagram is just as labour and startup cost intensive as running a server.
I'd even argue that servers are not as expensive. Have you seen how much money a good makeup kit costs? You can repurpose an old laptop or whatever as a server, or a raspberry pi... That's far cheaper than brushes for instance.
> and honestly I get that you picked makeup as some kind of example because you think that makeup is easy or something.
I think the parent picked makeup, not because they think it's easy, but it's a wildly different domain than system administration. Someone with a talent for makeup combined with a talent (or dogged determination) for creating mechanically good Instagram videos does not necessarily have a talent for system administration.
Even some sort of makeup-Instagram-sysadmin polymath doesn't necessarily have the time to put into each of those topics to actually make a video people want to watch, serve it well enough that a majority of the population can watch it no matter their device, and even get people to their server to even view the content. At some point you must have noticed the credits at the end of a TV show or movie. More than one person is often involved in the production so each one can focus on a particular task or even subtasks. You know, division of labor.
There's also the missing detail of no matter how good or a bad a makeup artist or Instagram video creator you are your makeup can't become a member of a malicious botnet or lock your home network down with ransomware if you do something wrong or leave it unattended for a weekend.
The "run your own server" meme essentially requires everyone interested in content creation of any sort to be independently wealthy polymaths with no outside responsibilities.
i don't think your parent was suggesting that youtube creators have the skills to run their own servers. i interpreted their post to mean that creating quality youtube content is a niche skill, similar to how running servers is a niche skill. i think they also were implying that, similar to the self-hosting demographic, a very small percentage of youtube users are high volume content creators.
For background, the author of this post wrote Manyverse, one of the most popular mobile SSB clients.
Implicit in this post, but never explicitly mentioned, is that people "not wanting to run servers" is actually mostly a side effect of poor design -- because running servers is for "technical people" who can put up with high cognitive load and lots of sharp edges, we don't design our server software to be easy to set up & free from unnecessary gotchas, and so we end up unnecessarily competing with other high-cognitive-load tasks these folks would like to do -- and the alternative is to make something like Manyverse, which is a client-server but that is no more complex to use than any other social media app unless you want to dig deeper.
(Similarly, "most people will never be able to / want to learn to code" makes a lot more sense in the context of a language like java, where most of the code a beginner must read & write is boilerplate with complex & dubious justifications, than in a language like python where for simple tasks there's a very close connection between the intended behavior & every piece of the implementation. Frontloading necessary learning attracts lore nerds & people who want to boast about their leet skillz, but scares off people who would like to just get something done -- so the more theory is necessary to use some particular stack, the more its user base fills up with ineffectual theory-wankers. This is a problem when theory has a dramatic benefit, but there's no good excuse for, ex., the stupid amount of manual configuration necessary to deploy a new apache -- where the "theory" rarely generalizes beyond the specifics of apache's own internals, & under most circumstances, reasonable defaults could easily be supplied or guessed.)
"Most people will never want to run servers" is, much like "most people will never want to own their own computers", mostly a statement about antihuman design tendencies & the way that particular groups have insulated themselves from them. People don't want their own computers so long as they take up an entire building and require dedicated air conditioning and card punches, and people don't want to run their own servers so long as deploying a new server involves dealing with IANA, NATs, port forwarding, DNS propagation, poking holes in firewalls, and other hassles. But a lot of people used to run napster off their PCs, and a lot of people still run bittorrent, despite all p2p software essentially being 'server software'.
People that want to run full ethereum nodes are a statistical anomaly. I feel that the author of this article wasted his time trying to pick a fight over semantics.
The key difference is that starting a Youtube channel is free and dead simple. The reason people don't want to run their own servers is because in most cases it's as expensive, or more expensive, than having someone else do it for you, and requires a huge amount of skill and patience, not to mention risk tolerance.
In general, I think this article misinterprets the statement that "People don’t want to run their own servers, and never will" by taking it too literally. The meaningful counter-argument isn't "this cannot be factually true, some people want to run their own servers", because that's clearly not what he meant. Obviously some people want to run their own servers, and some people have to run servers to keep the whole thing going, but that's a trivial point rather than a devastating counterargument.