Some people want to run their own servers | Hacker News Reader

Some people want to run their own servers

staltz.com

199 pointsnanomonkey145 comments

> 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.

I dont know what the article's definition for 'active creator' is, but

> requires a huge amount of skill and patience, not to mention risk tolerance.

making a good YouTube channel also requires those, and there's no financial incentive to small/beginner creators which are investing their time

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 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.

Yes, that's right and well worded.

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.

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.

> 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.

Doesn’t cut both ways.

> 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.

Same, basically my company said: “You’re title is now ‘devops engineer’ instead of ‘Systems Admin’”

Me: “Ah ok, do my responsibilities or pay change?”

Company: “No”

Nothing I do would I classify as “dev” work at all.

Tell all of that to your Mum when all she wants to do is buy an NFT...

You are not 99% of people, these are skills that seem trivial to you, everyone has different skills, very very far from everyone has sysadmin skills.

My mum doesn’t want to launch a website, if she did, she understands she’d have to learn. My mum is not so stupid she can’t learn things.

We don’t give people more freedom and control by telling them they’re incapable.

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 all very rough around the edges right now

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.

Dev is so easy theyre forcing sysadmins to do it these days too. See how silly that statement sounds?

> 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.

"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.

> Some people want to run their own servers

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.

If they’re using messaging apps, which most internet users are, they’re creating content.

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.

> Improved privacy and better control over owning your own content.

Ah yes, the increased privacy of an immutable public ledger of transactions shared globally. Much privacy, such control.

Did I mention crypto once? I'm talking about stuff like mastodon and scuttlebutt...

> 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.

> 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?

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"

> 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.

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.

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.

Also YouTube giveth, but sometimes YouTube taketh away.

This was my take too, it was clearly meant in general, not 100% of people. Taking it to mean 100% is a little disingenuous.

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.

> The reason people don't want to run their own servers is because in most cases it's as expensive.

This isn’t remotely true, AWS has a free tier of EC2 instances, not sure how much less expensive you can get.

> This isn’t remotely true, AWS has a free tier of EC2 instances, not sure how much less expensive you can get.

We’re talking about running your own server, not spinning up a virtual machine on someone else’s infrastructure.

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'.

> The crypto community has to ask themselves whether they want decentralization or money.

They already decided a long time ago. I don't know how the decision could be made any more clear.

In case you don't get what I'm alluding to: money. They chose the money. They keep choosing the money. They are very likely to always choose the money.

Almost everything in the crypto world with any traction is based on accumulating and flipping assets, or extracting value from those who are.

I think there's a lot of true innovation out there, which may one day serve as the core of the economy. But I think there's almost a bifurcation between work done that furthers that goal and work that is involved in the speculative frenzy. The latter is the vast majority of what's going on. From my vantage point of some distance from the scene, I don't see much overlap.

The part where it explains why "People don’t want to run their own servers, and never will." is a false statement because there are a few counter-examples seems like it's really missing the point. The vast majority of people don't want to, and indeed, most nerds also don't want to, even if you know some counterexamples.

This is one of the reasons I sometimes feel like commenting on the internet is like trying to walk through a minefield. Saying something as simple and obviously-hyperbolic as "everyone loves dogs!" can lead to a whole argument about how many people actually hate dogs, distracting from the original topic.

But it can actually get more annoying than that on the internet because accuracy and precision are things. So for example, mentioning a 1TB drive on the internet has the potential for comments to delve into a discussion about marketing and how actually it's not 1TB but XYZ GB.

It just feels exhausting.

I came to a similar conclusion recently.

Even if you are able to keep a discussion on track there are an infinite number of ways to view the world and it's possible to justify almost every perspective. Take any position on a topic and you'll find tons of blog posts, "studies", etc. that justify it. And the other person doesn't have the time or patience to actually look into what you are saying to see if it's legitimate.

great nick! --written on my own server

This is particularly true on HN, at least from what I’ve observed last year. On more than one occasion, I’ll type out a reply, only to close the window in frustration knowing where the conversation will eventually end up.

Good to know I'm not the only one that writes out replies and doesn't post them for this reason

Yes, it was incredibly clear what Moxie meant with that line (and it's been proven over and over) but some of the rounding error of people who like running their own stuff (I'm one of them) take this as a personal attack for some reason.

> The vast majority of people don't want to, and indeed, most nerds also don't want to, even if you know some counterexamples.

The point is that things are best designed in such a way that people may run their own servers but not that they must run their own servers.

The former is good even for the people who never intend to do it, because it implies that whoever is operating the servers you do use is in a competitive market that anyone can enter without having to overcome a network effect.

Sure, but Moxie's point was that current Blockchain protocols are designed for the latter: you either run your own server, or connect to some platform's server that is exposing whatever data they want over some other protocol. There is no client you can run that can talk to the distributed server network.

The problem is -- referring to Moxie's article here -- that some of the supposedly-decentralized things are already depending on or are fronted by specific centralized services, so even if you do decide to run your own server, it doesn't matter, because those centralized services have become de-facto gatekeepers to the decentralization in the background, and it's difficult (and sometimes impossible) for the average user to replace those.

It was very interesting to me to learn that if OpenSea de-lists your NFT, then it effectively ceases to exist. Sure, it's still on the blockchain, but any app that uses OpenSea's platform to list or operate on NFTs (which, turns out, is a lot of them right now) won't see it.

This might be fixable, but it's a big red flag to me that these problems already exist, with no proposed solutions.

I think you are not giving enough credit to the whole argument that André is making.

He is not saying

> See I found a few people who do run servers, therefore your argument is invalid.

His point is:

> It is correct that only a tiny percentage of people do run servers. But it is exactly this tiny percentage who make certain projects successful.

He then generalizes this observation. To paraphrase: "People don't want to run their own servers" is similar to "People don't want to write their online encylopedia" - true, but not the defining point.

In a sense this article doesn't say that the argument "people don't want to run their own servers" is wrong, but that it is the wrong argument.

GP got it right.

the whole article is fight over semantics that adds very little. i wanna hear about turning phones into servers, not the nick-picking. but yes, he wrote that what Moxie wrote was, and i quote, "factually wrong and sound like mere ideology".

But the point from the original Moxie article was petty clear: web3 protocols are designed in a way that is only distributed if everyone runs their own server. You can't access the Blockchain from a client (e.g. a mobile device). You can either run your own server, just for you; or, you can rely on the next Facebook or Amazon to run a node for you to connect to from your mobile device.

So, the fact that a few people can and will run their own servers doesn't help the vast majority who will not. Just like the fact that someone is running their own mail server will not allow me to have an email address: I can either run my ownail server, or use Gmail or maybe some smaller provider.

And in the past, the vast majority have chosen Gmail, because the email protocol didn't have a client app that could connect to the distributed mail network.

Similarly, the vast majority will choose, say, OpenSea if there is no client that can talk directly to the distributed NFT system in the Blockchain. And if everyone is using OpenSea, the Blockchain becomes irrelevant, just like the email standards are quasi-irrelevant for Gmail.

I don't think that's the case. From what I've read, the goal of this web3/crypto stuff is to bring decentralization to the masses. If only a subset of "nerds" want to run servers, that still doesn't cut it.

Yeah this attacks a straw man version of the original article. Obviously Moxie did not mean that literally nobody wants to run their own servers, just that one lesson from web2’s success is that very few do and as a result systems will tend towards centralization in the favor of the people who do run their own servers.

But now you're attacking a straw man version of TFA, which is saying "sure, hardly anyone wants to run their own servers, but enough people do, and that's all you need for a suitably distributed infrastructure".

Not the parent you're replying to, but I disagree with the assertion in your summary. I don't think the "enough people" here actually is enough. We're talking about a small fraction of a percent of all internet users. I don't see how that could be enough give the non-server-runners the touted privacy, security, censorship-resistance, and data-ownership benefits of decentralization.

I 100% agree, and said so elsewhere here. I'm just standing up for the claim(s) of TFA being stated and understood clearly.

>>> On Mastodon, 1 million active users but only 2 thousand (0.2%) instances On Tor, 2.5 millions users but only 6 thousand (0.24%) relay servers

The problem with this counterexample is that these are fringe platforms. Thriving in their own right, resilient, but a rounding error in chat and web traffic, respectively.

In other words, those who run their own servers are the fringe of a fringe.

The content creator comparison doesn’t make sense, with centralized hosting one content creator can supply an infinite amount of consumers. Not so with hosting.

Should also be added that Mastodon is fairly heavy weight and designed for many tenants. The way to get participate in the fediverse is no more to run a Mastodon instance anymore than the way to take part in IRC is to run an IRC relay. For better or worse, that's just not how it's designed to work.

This is in fact one of the main reasons I'm not on mastodon, because it's such a pain in the ass to set up, and I don't want to rent my identity on the Internet.

A stark contrast is something like Gemini, where the by far most common model is small self-hosted operations.

> The problem with this counterexample is that these are fringe platforms. Thriving in their own right, resilient, but a rounding error in chat and web traffic, respectively.

Which makes them the perfect place to incubate misconceptions that everybody loves hosting servers and decentralized platforms.

If all of your friends are doing it and talking about it and enthusiastic about it, it feels like everyone agrees with you. Yet these places are basically a tiny filter bubble, not an accurate cross-section of the internet or general public.

We're fixating a lot on "people don't want to run their own servers".

People don't care. They want to talk to their friends. Watch videos. See pictures of their grandkids. What they don't want is a lot of fiddly administration tasks with weird jargon they don't understand. Whether there's a server running on their computer or not is immaterial. If it's easy enough to use, it doesn't matter, as long as it lets them do the things they want to do.

Hopefully Web3 reaches the point that there's a 1-click installer, like downloading an app, that does everything you need it to do, so people can enjoy all of the things they want from the web with the benefits of decentralization.

"Download the Web3 app from Meta!"

That's a literally terrifying thought, yet most likely exactly how it'll play out in the end… :(

Running a server without having to go through fiddly administration tasks with weird jargon has eluded the software community for the past 50 years, or so. I don't know what web3 could do to fix this.

web3 isn't supposed to fix it.

people running their own servers would/might have given rise to a more decentralized system in which the putative problems web3 is claimed to fix would not have arisen.

I was responding to this:

> Hopefully Web3 reaches the point that there's a 1-click installer, like downloading an app, that does everything you need it to do, so people can enjoy all of the things they want from the web with the benefits of decentralization.

As long as web3 depends on everyone running their own server (or using someone else's server), i don't see why we would expect anything to change.

the "running their own server" thing is more retrospective, I think. Moxie's article was basically saying "IF people had ended up running their own servers from 94 onwards, we might not have gotten into this situation".

I don't think that is the point at all. The idea is: the only solution [given current/last gen web technologies like HTTP, email, IRC etc.] to have a decentralized web where most content is not in the hands of a small handful of mega corporations is for everyone to run their own web servers (literally everyone - each and every FB user would have to run their own HTTP server to achieve the same functionality; each and every gmail user would have to run their own email server). This is a major limitation of the internet protocols, and has lead to the current highly centralized web.

The promise of web 3 is to come up with new distributed protocols that allow people to have the same kind of functionality as today, but without having to cede control to FB/Google/MS etc. The reality of web 3 is probably closer to Moxie's article: these distributed protocols are actually closer to email infrastructure than to that promise, and the result will likely be similar to web 2: one or two major platforms will grow and offer a nice interface with the complex distributed network, and the vast majority of people will actually interact with the proprietary platform, not the underlying distributed protocol.

> I don't think that is the point at all.

Let's contrast my "summary":

> "IF people had ended up running their own servers from 94 onwards, we might not have gotten into this situation".

with your analysis:

> the only solution [ ... ] to have a decentralized web where most content is not in the hands of a small handful of mega corporations is for everyone to run their own web servers

Maybe we're reading this somehow very differently, but to me these looks at worst highly compatible points, and at best, the same point.

But I also agree with your summary in your 2nd paragraph.

> Maybe we're reading this somehow very differently, but to me these looks at worst highly compatible points, and at best, the same point.

I'm reading a nuance difference into it - the first statement, to me, sounds like it puts some blame on the people, while the second one (and, I believe, Moxie's article) puts the blame on the technologies that forced this situation.

Not sure I agree. What's another hacked-together shell script to a software engineer or system administrator? We've made servers manageable for ourselves, but rarely do we step outside of the bubble to ask how much of it is accidental complexity.

The networking layer isn't gonna change. That's given. Any change at the ISP level needs to be driven by policy decisions. But at the hardware/software-level? All of that is made up, and we could build much easier systems for network administration and app deployment.

Sorry for meta, but in my 30+ years in tech I haven't seen a technology that would spark so much discussion around whether it can be useful or not. People invest in interesting stuff, others build and roll those things out. The public finds it useful or not, that's how the world has been working for centuries.

I think at this point we are more in the psychological realm than technological one. This is a big sunk-cost-fallacy vs. sunk-cost-fallacy-recognition debate and it's getting very, very irritating already. Kudos to Moxie for giving probably the most clear technical explanation of what's going on behind the web3 marketing fluff, but the rest is getting very annoying.

Sorry, I had to let this out.

One major difference is the extent to which this scene is ensnaring regular people with FOMO and eye-popping numbers. Some people are seeing salvation, some people are seeing impending catastrophe, and some are just trying to get rich quick.

I think this is misreading the room.

Lots of people agree that there is a problem with data centralization, access and control that has arisen from of the most fundamental aspects of how "web2" works.

A few people are proposing that a specific technology, actually invented for something quite different, could be used to address some or all of these problems.

A lot more people are concerned that not only is the technology not well suited to the task of addressing these problems, but that seeking to use it for that purpose will actually make things worse, possibly catastrophically worse.

There were similar debates taking place online around the dawn of e-commerce. People concerned that pushing retail online would have dire impacts on physical stores, would upend local and national economies, would change the nature and range of things that we have access to in ways both small and large, good and bad.

Those debates did not get resolved, of course. Some idiots at Amazon went ahead and did it anyway, and here we are today, picking up after the storm.

Maybe view the discussion around the potential role of blockchain technology to address data centralization in a similar vein i.e. it will likely make no difference to what happens, but will at least leave behind a trace socio-archeological record that there was some disagreement before things changed.

> There were similar debates taking place online around the dawn of e-commerce.

I don't remember any debates in the 1990's about this. Some people built stuff, others just closed their physical shops and moved on to something else, or added online versions with delivery and survived. There was no debate per se, or if there was then it was by doing, not frigging talking.

Also, it didn't take Amazon 10 years to take over shopping. In 1999 Bezos himself famously said he's the most surprized person on Earth by how sucessful Amazon had become. It was obvious to most people e-commerce was there to change shopping forever, except for those who ran high street shops.

I was debating the e-commerce thing with people at UWash CS&E and Microsoft, including Mhyrvold, in 1993-1994. There were a few op eds and articles in magazines like the Atlantic, Harpers and New Republic that also touched on the matter. It wasn't a matter of dinner party discussion, but then neither is web3. It was somethng that those of us hanging around the edges of the coming e-commerce world were talking about on usenet.

I don't know where you got that story about Bezos saying that, but it's BS. Jeff was never surprised by the success of the company that he started. It also wasn't obvious at the start of amzn that it would work. Even Jeff had no idea the extent to which people would feel comfortable with the web as a platform, which is one reason why amzn had an early contigency implementation of an email-based store (yes, really).

Okay, here's Jeff Bezos literally saying that [1].

[1] https://edition.cnn.com/videos/business/2021/02/03/jeff-bezo...

Sure. It's still BS. Jeff never had any doubt after the end of 1995.

> So which one is it? Is “something that works” by definition also a force for censorship? Or is there actually a problem and a deep discomfort when your content is unilaterally taken down by the platform gods, no reason given and no right to dispute it?

I think this misrepresents Moxies point a bit. You can have quick iterations or a decentralized network with many heterogeneous clients, but the reality showed that both is nearly impossible. You can argue that the centralization at OpenSea is bad, but that's not the same as arguing that what OpenSea provides is bad. It's not a paradox.

It's the same as hating Google for slurping all our data and flooding the net with ads, while still appreciating it's great search functionality, fast free video hosting and best-in-class navigation offering.

> It's the same as hating Google for slurping all our data and flooding the net with ads, while still appreciating it's great search functionality, fast free video hosting and best-in-class navigation offering.

What makes it even worse is that one of the things that makes Google's navigation offering best-in-class is precisely that they've been slurping all our (motion) data.

This article starts by making a hilarious strawman. I clicked through to see if it was as hilarious as suspected, and it was. Which is great. Clearly Moxie did not mean “literally no single person wants to run a server” I mean come on. It is trivially true that “people” don’t want to run a server. It would be hard to generalize the category of “people” in a way to come to any other conclusion.

Exactly. Worst article ever. The writer is either very silly or completely disingenuous.

If a writer had to caveat every statement to be 100% accurate, they'd never get finished or if they did it would be so boring no one would read it. I want to caveat this with - some articles would get finished and some tiny percentage of them would actually get read. If you are still reading at this point, I guess I'm thankful for the caveat because it's now proven to be true. Let me caveat - "proven" is used in a non-scientific manner here because the above isn't strictly a proof. It's just a common sense usage of the word. Let me caveat "common": I'm speculating that it's common sense because it seems sensible to me and I feel like a common person.

My initial read on TFA was very positive. But then I read the comments here and I realized a substantial issue with the argument.

The author is precisely correct to point out that there are lots of arenas with large scale participation but only a tiny percentage of people "really making it happen".

However, many of the descriptions of the problem(s) that web3 is "intended to solve" would be left largely unaddressed by solutions that utilized TFA's observation.

Increasing (maybe?) the number of IT-minded people who run their own servers will not really do anything to address the centralization of data on the servers used by everybody else.

So in the end, I think his response to one of the main points of the referenced Moxie article isn't wrong, but it's a red herring.

If 5% of people want to run their own servers and 95% use AWS etc. then Moxies point still stands. These blog posts fee like some folks are trying to ride a popularity wave rather than make an actual argument.

What “running your own server” mostly fails at, in my view, is the constant needed maintenance. It doesn’t matter so much if running a server is initially difficult and time consuming. But the constant need to come back to it and fix bugs, install patches, implement security updates, add features yadayadayada … this makes it so perfect to be externalized as a service.

> These blog posts feel like some folks are trying to ride a popularity wave rather than make an actual argument.

Thank you. Precisely. Like, really? The guy said "nobody wants to run their own servers" but he actually meant "nobody [for all intents and purposes] wants to run their own servers"? Colour me pink!

> people don’t want to run servers

It’s true and I completely disagree with the take that we can’t generalize this statement because a few “nerds” still want to run their own. Generalizations are based on 99.9% of cases, and .1% are exceptions.

The truth is that most people don’t want to run their server, rather they can’t. Most users on the internet don’t have basic literacy, let alone knowing what a server is or how to run it. It is safe to assume most people don’t want to do much beyond one click access to whatever it is that they want to do on the internet. And the most efficient way of doing that is, you guessed it, platforms and centralization.

Somehow all of these articles choose to ignore what is for me the most critical point of web3: The fact that interacting with the blockchain costs real money.

If HN was a web3 app, then posting a comment would cost you money. I can't find anywhere how web3-enthusiasts justify this or what are the potential improvements around this.

EDIT: After some small research it looks like this is an Ethereum thing and that other platforms differ sligthly. But still seems to be a considerable drawback compared to the current web.

There is one silver lining here : spam being priced out (while hopefully still staying cheap enough for everyone to use).

Note how the hassle of dealing with spam is one of the aspects pushing centralization - also it's getting kind of scary how well spambots these days can just pass for a somewhat clueless user...

> There is one silver lining here : spam being priced out (while hopefully still staying cheap enough for everyone to use).

This may be true but so may be the opposite. We are already spammed with ads everywhere and someone pays for them.

> If HN was a web3 app, then posting a comment would cost you money. [ ... ] But still seems to be a considerable drawback compared to the current web.

This elides the way the current web works, though.

On the one hand, you've got some places like HN, that are coupled to revenue generators of sufficient magnitude that they can (AFAICT on my uBlock-Origin'ed browser) run without ads.

And then you've got the rest of the web, which has taken the old ad-funded model of publishing and turned it up to 11.

So when we say "the current web" in a context like this, it's important to differentiate between which of these (at least two) segments you're referring to.

[ EDIT: granted, there are a few new models creeping around the edges now. For all of its faults, substack is a mostly previously unseen thing. ]

If I understood correctly, this comparison is wrong. The money GP talks about is needed to interact with the infrastructure (blockchain), not the content publisher. If HN were a web3 app, posting a comment would cost you money even if HN runs without ad, because as a web3 app it needs blockchain interactions that cost money.

When I first read about Web3, I thought the exact same thing. The fact that there are fees to deploy and run dapps is right there in the design. And yes, it does seem like it's the intended design to have users pay the ether to do so.

But that requires users to download and learn how to use a dapp browser, and to front the ether for running them. That's going to be a non-starter, because with the possible exception of a very small number of people, nobody wants to do any of those things.

But hypothetically, an app developer could developer their backend as a dapp, and a HTML/CSS/Javascript frontend, essentially using the blockchain as a replacement for AWS. Is this useful? I suspect it isn't. One could argue that blockchains have better uptime than AWS, but I bet the price would be higher and the performance lower. The only other benefits you get are that anyone can audit the dapp (if they trust your Web server to actually be using it), and that you're no longer beholden to a socially-destructive giant company like Amazon.

As Moxie's original article pointed out though, you'd still be beholden to the few companies or individuals that run etherium nodes. This is, again, something that approximately nobody wants to do, and if Web3 ever took off, I suspect the bandwidth and storage requirements of running a node would increase to the point where only socially-destructive giant companies could afford to do so.

This is one of those "technically correct" statements that is mostly useless.

Yes, some people do want to run their own servers. The vast majority -- more than 99%, by the article's own admission -- do not. To me, "people don't want to run their own servers, and never will" is completely accurate.

I am a "nerd" by Moxie's definition, and I mostly do not want to run my own servers. I have a VPS[0] that I use as a playground, and to run some useful but non-critical things, but it doesn't run anything remotely critical, and if it went down for hours or days, it wouldn't be a big deal.

The idea of running my own infra that's critical or at least important to my participation in digital matters does not interest me at all. For a "web0" example, I absolutely would not want to run my own email server. (I did this when I was much younger, and regretted doing so.)

I also disagree that the small, sub-1% faction that will run servers is enough for this web3 decentralization-for-the-masses world. Ultimately someone else is controlling your traffic and your content, and there are a limited number of others you can switch to if you don't like the current steward you've chosen.

[0] And also consider that having a VPS is only partially running my own server. I'm running the OS and applications on it (or, rather, on a VM running on that server), but someone else makes sure the server hardware and network are functioning, and is on the hook to fix them if they aren't.

>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 "promise of web3" has been that there are no centralized giants. The problem isn't one of motivation to be a YouTube creator, it's infrastructure. The cost to becoming a YouTube creator is 0 due to centralization. The cost of building YouTube is millions of dollars. Even in the YouTube case - people don't want to run their servers.

Secondly, Mastodon and Tor do not have mainstream appeal, nor are they platforms on which other economies are built. Their comparison to YouTube is absurd.

The only place I see people talking about web3 is HN and a few random sparsely commented articles on /.

No one I work with has any plans to learn about it or use it.

It's the exact opposite of how web 2.0 was.

I feel like this is a pretty thin strawman attack. Moxie's core point wasn't that people don't want to run servers, it was that most of the Web3 stuff is heavily server-focused and very little effort has gone into passing these benefits down to the actual users' clients, and the work you have to do on top of a server to make an acceptable UX leads to a centralized to the type of lock-in that Web3 is purportedly about solving.

Quibbling over what percentage of people are willing to run a server seems like a bad hill to die on—the reason Moxie emphasized it was because it should be self-evidently true at this point with the exceptions proving the rule as they say.

People dont want to eat all 3 of the following dishes on the same night [Lutefisk, Durian, Casu_martzu]

There exists at least 1 person who does. So at least 1.3×10^(-8)% people like doing it.

Does that warrant: "Most People", "A Majority", etc?

Speech needs to be accurate, but not always pedantic.

"People like to watch sports"

I dont. I find it a huge waste of time. I am in a tiny minority. (if you include all sports)

People except for [ThinkBeat, Bob, Ermia, Ashi] like to watch sports.

https://en.wikipedia.org/wiki/Lutefisk https://en.wikipedia.org/wiki/Durian https://en.wikipedia.org/wiki/Casu_martzu

Kind of impressed by how civil the discourse has been, both moxie's original post and this response.

Really not the norm when crypto is being discussed

I agree. But the post of Moxie was also the only one till now that actually showed what most of us think.

But way better explained, done and researched than anyone tried. Because most don't even try.

I think it leveled both sides ( including me).

Interpreting an obvious generalization as absolute fact, is not conducive to a productive discussion.

Just replace "own servers" with "own restaurants" and understand the original meaning:

> Thus saying “People don’t want to run their own restaurants” is akin to saying “People don’t want to have their own kitchen”. Both sentences contain the same amount of statistical bullshit.

oh wow. did we read the same article?

these criticisms sound odd to me. the whole section about OpenSea removing Moxie's NFT. and how he wrote that he wasn't out to complain about them, is not how i read it at all. this is so odd.

i recommend people read Moxie's article.

the comment about people not wanting to run their own servers resonated with me. and with many i suppose. i mean, that's what we have been doing for a while now. that's why the cloud is thriving. that's why we like managed services.

i would not have added "and never will" though ¯\_(ツ)_/¯ what i want is the author to elaborate on how to turn phones into servers. i want that.

With a very, very few exceptions, I think "people don't want to run their own servers and never will" is true. They don't want to run servers, but they will run a server if the only way to get what they want is to run a server.

There are always a few people who want to run a server and it's not just a means to an end. But they are so incredibly few they aren't worth taking into consideration.

Put another way: All else being equal, if you give someone a choice to run a server or not, they won't.

Let's say I want to get a little RPI that comes with a small LED screen and a button to generate a five-digit initial login code. Then you connect over wifi through your phone can select from a dropdown menu what you want your home server to do. Let's even say that the only entry is something ridiculously easy like "Quote of the Day."

The only other requirement-- the box gets automatically updated like a Chromebook does.

Are there any extant FOSS home server projects that can achieve this minimum level of usability and reliability? If so I'd like to know about them. If not then Moxie wins.

As someone who wrote their own "Quote of the Day" server [1] I will agree that the QOTD protocol (RFC-865/STD-23) is simple, but not quite the "ridiculously easy" as you put it. Let's assume we have something like that. The code for the service itself is pretty simple (listen for a connection to TCP port 17, send a quote, close the connection) so there shouldn't be that many updates for the QOTD program. But can the user add new quotes? If so, how secure it is? Can that be exploited? How often should the quote be changed? Every request? Once a day? Something in between? How is it configured? Do you need to document DNS settings? Are there mitigations to keep a client from swamping the service with requests? [2] Are requests logged (say, to monitor bandwidth usage or to see how popular a service it is)? Where? How long?

Nothing is ever "easy."

[1] qotd.conman.org

[2] Not a problem I've had yet.

I would love to see servers become easier to host. Personally I enjoy hosting my own services, its super rewarding and I've learned an absolute ton about computers and networking.

I think making it easy to host reliable and secure servers we would see more people jumping in. Maybe not huge amounts but enough to change the ecosystem.

> I've learned an absolute ton about computers and networking.

I think that’s the problem. Most people’s goal is not to learn a ton about computers and networking, it’s to accomplish some other task. These days rarely is running your own server the easiest way to accomplish most tasks, even if it is an option. It’s usually easier to use someone else’s server, and into running your own is easier than using someone else’s, most people will continue to do use someone else’s.

In my opinion, learning this stuff enables you to accomplish other tasks easier and more effectively. Since learning about this stuff, I can now stand up any self hosted services that I want with relative ease. That is the advantage to deeper learning in tech

On top of this, I also better understand the privacy implications of using other people's services. It helps me operate in our current tech ecosystem far more effectively

>I would love to see servers become easier to host.

Unfortunately blockchain, which is what people are trying to push as part of the decentralization, is making servers harder to host.

You can run a Bittorrent server or an IPFS node on a $50 Raspberry Pi. But if you want to spin up a Filecoin node, you need a server that has at least 8 CPU cores, at least 32 GB of RAM and enough SSD space to contain the chain that's expanding at a rate of 38 GiB per day. The listed minimum requirements for the mining server are even worse - at least 8 CPU cores, at least 128 GB of RAM and a good chunk of NVMe SSD (1 TB preferably), with a recommendation for a "powerful GPU" to boot. The only sensible option is the Lite node that can run on a measly dual-core with 2 GB of RAM - which can only operate on such low hardware requirements because the lite nodes do not actually contain the blockchain.

I'm glad to see this post, as it says exactly what I thought when I read the referenced sections in Moxie's post. I know I live in a particular echo chamber but owning two ISPs has meant a sizeable proportion of my Dunbar's Number are those who gladly host their own servers, run their own infrastructures, and decline the ongoing consolidation of the Internet. We generally consider being part of The Internet for example to mean being part of the DFZ (Default Free Zone), and anything less is to be a consumer. I know we are the extreme minority of users, but it's users like us who keep the world connected and accessible. Without people running their own ASNs, hosting their own infrastructures, etc, we would all be reliant on a small (relatively) number of huge scale businesses. And these users permit disruptive Internet infrastructure businesses to exist.

Take for example the rural WISPs in the US or South and Eastern Europe who provide coverage for huge quantities of users in otherwise unserved locations. Or the organisations running Wikipedia and The Internet Archive. Hosters like OVH, Hetzner, even Digital Ocean. Broadcasters like BBC and Netflix. Fibre providers like B4RN and even new ISP models like Starlink.

If people didn't want to run their own infrastructure the underpinnings of the Internet would become stagnant and opaque owned by a small quantity of hyper scale businesses. This isn't the Internet promise that I grew up through.

This is entirely missing the point of Moxie's original article. The promise of web1 and web3 was that people like you wouldn't be needed at all - it would be easy for literally everyone to participate in the distributed internet and run their own stuff, just like people run their own browsers today.

Web2 quickly showed that web1's promise was false: to participate in the distributed network, you had to run a server of your own, or use someone else's server - and this is what the overwhelming majority of people did - at first, the gigantic masses of amateurs, and finally even most professionals. Anyone can publish a picture on the open WWW: just get a DNS name, a static IP, a TLS certificate, and run your own HTTPS server, then simply give your friends and anyone else who might be interested the address over email. Oh, you'll have to run your own email server as well, but that's easier now that you have a DNS name and certificates and all that - don't forget about a few simple things like DKIM and whatever else is required, set up some spam filters, and you're golden. You shouldn't need to spend more than a few days every month making sure everything is up to date, reading logs for signs of trouble, and maybe reading the occasional forum or five for troubleshooting.

Or, you can make an Instagram account, give them all the data they ask for, and post a picture. You may want to change your password once a year, and make sure your Gmail app is configured for safe 2fa.

Moxie's point was that the same thing is currently happening with web3: interacting with the base networks directly is complex and requires larger investments, and it's much easier to interact with a nice service run by someone else on a proprietary platform that spares you all this trouble.

Reiterating the fundamentals of Moxie's post doesn't change anything, and I agree with the way you've described it, except you've missed MY point entirely.

I'm not trying to be elitist and deny access to everyone to contribute, but there is a vast difference in outcome between your examples you gave.

In no way would I ever propose that if someone wants to post a picture (the example you gave) that they should go to all the efforts you described hosting their own server for that purpose. That's ridiculous. That problem is well solved with Web2.

What I'm saying is that for the Internet to exist you need people prepared to get their hands dirty (sometimes quite literally) and build low level infrastructure.

Instagram and other platforms can't get fibre in to people's homes, or backhaul cities together, or get signal to your mobile phone.

> What I'm saying is that for the Internet to exist you need people prepared to get their hands dirty (sometimes quite literally) and build low level infrastructure.

I think no one is contesting this.

> That problem is well solved with Web2.

Maybe this is the major disagreement? Many are dissatisfied with Web 2 because it has concentrated the web, for the vast majority of people, in the hands of 2-5 platforms owned by major corporations (Facebook and all its apps, Google and YouTube, Microsoft's Office 365).

The promise of web 3 is a decentralized Web - same capabilities, but provided by distributed apps run and financed by blockchain techs. This is the promise that Moxie likes, but that hasn't matched his actual experience with the actual technologies being implemented - which seem to resemble Web 1 / Web 2 much more than the promise of Web 3.

> The crypto community has to ask themselves whether they want decentralization or money. Sometimes they can have both, but at some point, they’ll be forced to make a critical choice between one or the other, and that’s how we can know what is the primary value upheld by the community.

I've seen this sentiment quite a bit and it bugs me because you cannot easily separate money from the system.

When choosing between centralized, federated, and decentralized services you have the same fundamental problem.

Do you:

A) Host it yourself or

B) Have someone else host it for you.

A is a non-starter for most people. However, B introduces a new problem, the resource allocation problem. We live in a world with finite resources and we need to figure out how to allocate them. Money is a solution to the resource allocation problem.

Decentralized and federated software is great but it doesn't solve the whole problem. This is one of the reasons federated services haven't taken off yet and why crypto seems to have some traction.

A great example is IPFS. It's an interesting piece of software, however, you still need people to host files on IPFS. This takes resources and without money you rely on people hosting files out of the goodness of their hearts. Unfortunately this isn't a reliable motivator which is why Filecoin exists.

If everything is truly decentralized and there are tons of options, the net value in those options will approach zero because they're all interchangeable. Thus, I agree that decentralization and money cannot be the same goal.

Blockchains actually are decentralized.

Exchanges are centralized but even then there's dozens to choose from.

Plenty of people exchange crypto for goods and services directly without converting to fiat.

It's almost like this argument is talking past itself and ignoring the reality on the ground.

I know someone who makes her own dyes to color her own yarn for knitting.

And yet with vanishingly few exceptions, not even the most hardcore crafters want to dye their own yarn.

Thus it is for servers.

When you observed the world, and found that it is somewhat centralized, to an extend that you'd like to see a change.

Have you figured out why it is centralized in the first place?

Just trying to re-start the process and expect to see a difference? It's going to end the same way.

Because I'm sure centralization is not a Web2.0 property, but a human property. You don't change Web2 to do decentralization, you change humanity.

I think these two are talking past each other a bit because they're ignoring the fact that running your own server is still a complete pain. Any technical or social concept that could possibly require mass adoption of self-hosting is just moot until it as easy as: 1) plug in device to wall 2) connect to wi-fi 3) login once 4) start installing services 5) never worry about security or backups again. You could even mail out the devices to bootstrap your network, BankAmericard-in-Fresno style.

There are people that would run their own server on principle, and there are many, many more that would do it if you had to do in order to get on a cool social network with all your friends and top celebrities or whatever. I want to run my own server, but I don't because: -I don't want my data unencrypted on a cloud VPS

-I don't know a good dynamic DNS provider off the top of my head

-I don't want to host services on my primary desktop unsandboxed and it takes up too much power

-If I use another device I would need to: 1) segment my LAN, which my ISP router doesn't seem to support 2) manually configure updates, backups, port knocking, etc. 3) string cables around or otherwise find some hidden or unaesthetic place to put it

> It’s compelling to look at the non-hosting users, conclude that they are a smashing majority (99.8% !!!!), and make a sweeping conclusion that no one wants to host their own servers. Yet it’s precisely because of those 0.2% that the system is capable of existing. Take them away, and the whole system dies.

The Matrix chat protocol is like this - you have many more users than servers, but you can switch servers, the servers can talk to each other (federated), and there are even multiple implementations of server software.

I've moved almost 100 friends (in their 20s) onto Matrix - Element is getting really good, and some use FluffyChat or SchildiChat. They aren't running their own servers, but they can if they want (unlike Signal), or they can switch down the track.

It's pretty easy to host your own - it took me an hour or two to set a Matrix server up on a $5/month VPS, and maintenance involves running one setup command again: https://github.com/spantaleev/matrix-docker-ansible-deploy

1. How about we stop calling this NFT thing Web3? Isn't it in its current form at least an insult to both the original web and web2 (and with that I mean what it was thought of back when we talked about it: interactive pages leading to user generated conten: everything from comments to ratings to tagging/folksonomy etc)

Shouldn't Web3 be reserved for something important and web related, not used to hype up the current crypto rush which seems to be mostly scams?

To be fair and honest I should note that I was wrong about the web when it was introduced around here when I was an early teenager and I won't get caught in that mistake again so I assume something useful will come out of todays crypto hype. (Again no, I refuse to call it web3.)

Although to be fair to teenage self what I doubted most was the ads about the web and in that regard I was mostly right. And if someone comes around 25 years from now and points to this comment: No, I'm not saying distributed crypto systems aren't useful, only that the current ones are deeply flawed: Some NFTs at this point is just a URL reference to another webserver that can be taken offline at any point! Others are run by commercial companies in ways that mean they can disappear at a whim. There is no permanent ownership in most of todays systems.

2. For some reason people around here think that if something doesn't have a significant percentage of the users it doesn't matter. So since less than 5 or even less than 1 percent self host it doesn't matter. Wrong! Just remember: at some point Google had less than 1% of the worlds searches. Linux was a thing for research communities and nerdy teenagers.

It takes just a few to keep the seeds of knowledge and passion alive and if and when the time comes it suddenly breaks through in one form or another.

I want to add my two cents in that "run their own server" does not necessarily mean hosting open web exposed services. I started slowly in 2017 and have currently 4 family members' houses connected via IPSEC (one of them cross-atlantic). All internal information exchange, hosting private services etc.

You can argue that for IPSEC (or VPN or Wireguard) you still need _some_ port open, but getting that single port secure is much more hassle free than keeping control of (e.g.) 20 individual web services. And most of it is already set by default. This provides already a lot of benefits: I can try out technology, I am (largely) independent and I can migrate on my own terms and schedule, not when forced by cloud providers. I still have external "hosting", but it is very basic and controlled through CI on my own private Gitlab ( pushing static sites to the public through SSH).

I am not sure whether this is fully compatible with web3/crypto, but to me it was the first step in getting some control back and it feels really good, despite the work.

“Protocol changes more slowly than a platform” is really the crux of the entire issue here but not stated as such. The key advantage of centralized services is that they are fast, coordinated, more intelligent and also easier to love than a mob. The only things that a mob will ever do are things that don’t require you to be smart, coordinated or fast. It’s great for some things but here’s the catch: nobody is going to invest money into a mob because there’s no leader and no way to be confident about your investment. People are going to gush to their friends about “that mob” the same way they might have with Instagram.

we’ve just spent an enormous amount of time and money learning the very simple and obvious fact that in order to have nice things we must have smart, honorable men around who can lead our society to victory at the helm of these centralized businesses. They think if they fiddle around with decentralized stuff enough they can eliminate the need for exceptional business leaders.

Summary:

1. healthy decentralized services often have <1% of users who do "want to run a server," and the whole thing works as a result 2. immutability / inability to change quickly is good for protocols 3. crypto community does need to decide whether it cares about money or decentralization

Took a while getting to the good part of 1. Should have skipped "'no one wants to run a server' is factually incorrect!"

I loved the Moxie piece but it was strange bc Moxie is a legendary cryptopunk who presumably believes decentralization is good (though hard to attain/maintain)... right? Moxie made excellent points but left me wondering what he actually thinks, or thinks should happen.

I also liked the pushback here about email being an example of failed decentralization, because Gmail is big or something? Email seems like the last major decentralized communication protocol still standing, despite incredible centralization/"balkanization" of the messaging space.

You can hustle your way to financial independence on a new shiny like Web 3, even as its gets harder to do the same on Web 2. If the only way to win is to be innovative and rich, and the only way to be rich is to ride a gold rush, then by god we’ll make a whole new industry to attract investment to fund said gold rush. This is my “bullshit industries” thesis - an economy wide manifestation of bullshit jobs.

Maybe this cyclic destruction and reinvention is one of the ways we know of to keep rent seeking in check, enable social mobility and the sustain the illusion of progress. True innovation does all that and more, but it really is very difficult, slow and unpredictable. Not at all like Amazon Prime (TM) 1 day delivery. 2/5 stars.

My plan is to gradually add more stories, diversions, pictures, smart words and a few controversies into this comment until it grows into a series of substack posts and eventually a book i can sell.

I really appreciate you putting this out there. Doubly so with the respectful treatment of it. And for the record my take is very similar to Moxies. But, many comments so far have attempted to refute your critique, fundamentally, on a point of scale (i.e. 100% of Mastadon is a rounding error to overall messaging traffic, so the percent that run servers is doubly irrelevant).

From a VC backed point-of-view that is a legitimate point of push back at this stage. From a technology leader point of view, it is not, exactly. Early adopters have _something_ to adopt early, because of these fringe groups. This is where creativity starts and eventually meets engineering and productization. Some things do shake out to 90/9/1 for _all_ internet traffic in a certain category. Some don't. I think pressures also ebb and flow. I remeber when Googles moniker of "don't be evil" wasn't an ironic, bitter joke.

Running a server today requires someone to watch it, and fix things when it breaks, or gets hacked. They always get hacked. It's simply not possible to put a device on the internet, and have years or decades of uptime, as it is off the internet. There is no operating system available that is suitable for the task.

> The people at the end of the line who are flipping NFTs do not fundamentally care about distributed trust models or payment mechanics, but they care about where the money is.

This is ultimately the issue, some people want to grown their own vegetables in their backyard, or chickens, but they do it for their own purposes, the loud majority in crypto space(now pushing the web3 branding because it sounds more revolutionary) is in it for the money, and since they lack the tech knowledge to build stuff around crypto they focus on the marketing/virality side, because that will make money faster.

The irony starts when what these people want is to control the supply and distribution channels for what is supposed to be a decentralized set of systems. I think the comparison to the fediverse is very good, people are in it for the tech, no one is blindly pushing it to increase the value of their virtual assets.

> some people want to run their own servers.

I've been involved in the blockchain space for several years and I think this is absolutely critical.

The entire decentralization and security aspect depends on people running their own servers. While this shouldn't be imposed on regular users, it's unfathomable to think that a big investor or a big finance firm would rely on a server hosted by some third party... If you're transferring millions of dollars worth of tokens on a blockchain, why would you take on the third-party risk of trusting someone else's node/API when you can remove that risk entirely by running your own server? There is a certain threshold of value above which you absolutely do want to be running your own node and it would be foolish not to.

IMO, this is one of my biggest concerns about Ethereum. It's very difficult to run your own node.

> If you're transferring millions of dollars worth of tokens on a blockchain, why would you take on the third-party risk of trusting someone else's node/API when you can remove that risk entirely by running your own server

Or, you know, skip blockchain entirely and use actual trust mechanisms that already exist in the world.

This is about static IP addresses, not "servers". Anyone would be happy to run a server on their cellphone, but there's no good way to publish the address.

It's DNS that makes us think a static IP is needed to have a "server". When there's an alternative, this will change

I understand this issue with cellular end-points. I saw a comment on one of the two previous posts here, where a user complained it was impossible to obtain static IP allocations for setting up their own "servers". I found this odd, since in my own history this has never been a problem for a nominal fee, even when I had a 9600Baud modem connecting my network with PPP. I am now wondering, if perhaps I've misinterpreted the concern, that people would like to be able to serve content from their smart phones with fixed IPv4 addressing? What is the actual use case for this?

I have comcast internet with my own modem/router. Comcast will not give me a static IP unless I use their rental modem/router, which costs quite a bit monthly. I use ddns to work around this issue, and it mostly works fine.

I agree with this. Comcast, is certainly not the cheapest option. We pay right now around $220/mo to have our /29 addresses routed through Comcast. It was about half that over bonded channels of VDSL. But these rates may be a non-starter for a home setup. But it still seems pretty low cost, given what I hear from most of my friends about what they're paying for Cellular plus Home-entertainment on a monthly basis.

I also don't have a static ip at home, but have a hub-star private network that runs over Cisco routers over the Comcast network that allows multiple employees to share in the resources, from the single connection with public facing IPs, with pretty low latency through the private network. It's all done on the cheap, but fast enough I can remote into the office and work on 3d models or photoshop. Certainly fast enough for routing side projects for testing. Obviously this is more complex and costly than the initial home setup of a single node. But even in more rural situations it should be easy to build up networks of like minded people with local ISPs which could share resources like this. Apply for /24, get it setup with an ISP, and then VPN the rest of the traffic in from fellow customers.

Ah, that's a good point ! While getting a static IP prefix subnet should be the default for regular connections, and the situation for that should only get better with time, AFAIK cellular Internet is still kind of an unsolved problem... since moving to another cell means getting another prefix !

But then isn't this the kind of issues that can also be handled by decentralization via replication of data across the Internet ? (Would probably still be prohibitive for data that is both obscure and heavy...)

Some countries want to run their own currencies, but most of them really just outsource to the dollar or a printing press in Germany.

Some countries want to run their own militaries, but most of them really just outsource to the us defense contractors.

Some countries want to run their own languages, or religion, but most of them just outsource to English or Spanish, Christianity or Islam.

Some countries want to run their own operating systems, but most of them just outsource to microsoft.

Some countries want to run their own fans, but most just outsource to Asml and TSmc.

It’s almost like we already have a global order.

I run my own server for myself, which is just that I have multiple machines. And I like them to be accessible by me no matter where I am.

The related article/thread from yesterday:

My First Impressions of Web3 - https://news.ycombinator.com/item?id=29845208 - Jan 2022 (1049 comments)

My mother used to say “you’re being obtuse”. It drove me up the wall (but I have to admit she had a point most of the time).

She meant I was ignoring or missing the obvious (implying I should know better — mostly true — and that I did it intentionally — usually not true, sadly, but in retrospect it’s nice she had more respect for my intelligence than I probably deserved).

Maybe “people don’t want to run their own servers” doesn’t mean literally not a single individual wants to run their own servers? But rather an insignificant fraction of people want to run their own servers?

Staltz, you’re being obtuse.

"Hosting a server is firmly a very rare decision that users take, yet the whole system depends entirely on that minority of rare individuals. "

No, not for most systems.

Using 'Web 1' tech, anyone can run a server today, they always could, it's not that hard.

Web 1 provide the basis for most of the kinds of decentralization that Web 3 proponents seem to think will come from some new approach.

It's not perfect, but it's still probably better than most Web 3 approaches such as blockchain etc..

me: No one wants to ride around in a horse and buggy

stalz: Oh yes they do, I can assure you that the Amish ride around in a horse and buggy

me: how is taking things literally and pretending to miss the point helping this conversation?

TL;DR the whole article is this section:

“”” “People don’t want to run their own servers, and never will.”

This cannot be factually true, some people want to run their own servers. “””

That’s such a boring response to me. If you’re like me, skip the article. It’s basically just that stuff.

Reply on news.ycombinator.com