The PeerTube content bootstrap fund | Hacker News Reader

The PeerTube content bootstrap fund

sourcehut.org

189 pointsddevault77 comments

I understand that PeerTube is an interesting project but every time I look into it it seems hostile to any kind of "normal" user so I'm just assuming that's not the goal of the project to become some kind of popular alternative?

I tried to pretend I'm a normal user and looked at the website and right now it goes like this:

Let's say you are interested in the project, maybe you heard that it's a more open alternative to YouTube and you are intrigued. So you end up on: https://joinpeertube.org

There's no content visible or anything that would spark any interest in a person looking for a video site. I realized that joinpeertube.org is the site of the open source project and I have to look at a specific instance to get a more "YouTube"-like experience. So I click on "Instance list" which brings me to:

https://joinpeertube.org/instances#instances-list

The first entry in the instance list is "No Censorship Tube" and you don't really see if it's an active instance, popular or anything as it just says "Follows 0 instances" which makes it seem like an abandoned project.

Am I using it wrong?

In my opinion, the main problem here is that PeerTube is more expensive to host than, for example, Mastodon or Pleroma. The model is not going to be well-suited to drive-by creators hoping to sign up and start uploading in 5 minutes, because the cost of hosting that user is pretty high for the instance administrator.

With this initiative, I hope to start bootstrapping instances which are comparible to cooperatives. When money and content start flowing into the ecosystem, I hope it'll push for more development towards improving the user experience for non-creators, and provide a bunch of good content to incentivize federation. I hope to see more instances (and better upstream support for instances) which are not focused on uploading/creating - but on consuming - then make this the path that most users follow to participate in the network.

It would be really useful that some instance managers out there give us a cost breakdown on how much it takes to run a "regular" instance with specific volume/traffic numbers and videos served. Transparency can help people to start in case the cost is lower than what they expected.

As a rough estimate:

1 TB cost about $1.20

HD is 5MB per second, UHD is 25MB per second. [1]

If you stream a 3min video in HD, that's about 200seconds which lead to 1GB traffic.

So you can stream 1000 videos for $1.20, or one video for 0.12 cents.

[1]https://www.epiphan.com/blog/bandwidth-for-streaming/

That’s megabits per second, not megabytes, so your numbers are off by a factor of 8.

Where has that $1.20 per terabyte figure come from? It doesn’t match the sorts of rates that I’ve seen talked about. As an example of commercial retail (not wholesale) rates: Backblaze B2 cloud storage is one of the better options available, at $5/TB/month for storage and $10/TB for bandwidth per https://www.backblaze.com/b2/cloud-storage-pricing.html. (AWS, Azure and Google Cloud are all at least five times the cost for traffic. Seriously.) Meanwhile, your standard $5/month VPS (Digital Ocean, Vultr, &c.) tends to include 1TB of traffic per month (and scale up at that rate for a bit, though not far), but charge overages at $10/TB. I have no idea what’s going on with that lot, because Digital Ocean does bandwidth pooling across all your droplets, so it’ll be cheaper to add a new e.g. $20/month droplet and get 4TB of bandwidth, rather than paying $40 for that 4TB of bandwidth.

But remember with PeerTube that the idea is that some of your bandwidth is covered by peering. I have no idea at all what the expected fractions are, but it’s going to depend heavily on your workload: videos just released that everyone’s watching right now will get better peering than videos from some time ago with scattered viewing (where I would expect you would get just minimal peering effects, but my opinion is hunch only and lacks any actual knowledge).

5Mbps is on the high side for HD (1920×1080 @ 30fps) video streaming. I’d think 3.5Mbps was a more typical figure that people would speak of, and for many sorts of content even 2.5Mbps can be more than you need.

3.5Mbps for three minutes is less than 80MB, rather than your 5MBps’s 1GB. At B2 rates (which are drastically higher than what you quoted), each of your three minute videos costs 0.08 cents to view, and each hour of video costs about one and a half of your pennies of the United States of America.

We come to different results because all my Bs were indeed bits, not bytes. I would love to be confirmed that B2 calculates download costs in bytes and not bits, but I assume that the reference dimension is bits since that would match the AWS numbers. So that $10/TB could actually be $80/TByte.

Some time ago I asked about cheapest rates: [1]

>I'm starting an ISP again, we will charge around $0,25 per TB

The cheapest regularly available rate of which I am aware is hetzner.de at $1.20 (or rather euro) [4].

Amazon is at $0.09 per GB ($0.05 after 150TB). That's $90 per TB [2],[3]. You have to use Cloudfront to further reduce that cost, much like with Backblaze. But Cloudfront doesn't allow you to stream videos or serve pictures.

In both cases I assume that it is TBit since it is network traffic. I haven't found any further clarification so I assume that they quote the more expensive option.

That 80MByte video would be 640MBit. That's not too far off the 1GBit.

[1] https://news.ycombinator.com/item?id=22760447

[2] https://aws.amazon.com/ec2/pricing/on-demand/

[3] https://aws.amazon.com/s3/pricing/

[4] Some providers offer a rate-limited interface, e.g. 100Mbit, with unlimited traffic. This could lead to 32TByte per month for something like $10. However, like Cloudfront, I assume that the provider has the option to end the contract, otherwise there would be plenty of cheap image hosters.

> We come to different results because all my Bs were bits, not bytes.

Such Bs are usually bs, e.g. 'Mbps' vs. 'MB/s'.

So the GB on the AWS sites are actually bytes?

Excepting errors, capital B is always bytes, and lowercase b is always bits. Transfer rates are always done in bits (5Mbps and the likes), but amounts are always done in bytes (5MB and the likes).

The B2 rates are $10/TB. Yes, ⅑–⅕ of the price of AWS, because AWS data transfer is outrageously expensive. (Refer to the B2 pricing page again, it shows this comparison.)

Thanks.

The prices are even more surprising if that quote by morphle is true and he can offer a TB for $0.25. From that perspective, even B2 is expensive.

At the level of colocation and the likes, I believe it’s standard to just pay for the pipe and you can use it as heavily as you like. Some hosting providers extend that to their users, as well.

But there are practicality concerns to this for the specific domain of video distribution; suppose you’re a solo video producer: most of the time there’s a certain background level of watching, but when you release a new video there’s a massive spike of watchers. So in this case, having an enormously fat pipe and charging by data transferred is probably better for you than having a thin pipe that you can use as heavily as you like. If you have a 1Gbps link, for example, and 5Mbps video, you can never have even 200 people watching at any given time. But if you have a 40Gbps link and are charged by data transferred, you’ll be able to cope with a surge of closer to 8,000 viewers at once. (I bet that in practice it’d be struggling before even half that, but the point stands.)

I know some hosting colocation deals are done as a certain baseline with burst capacity, e.g. 1Gbps with 10Gbps burst capacity, which would mean something like “you can use up to 10Gbps, so long as 99% of the time you stay below 1Gbps”.

But all of these sorts of considerations are mostly important only when you’re a big shot, not when you’re just starting out or if you’re just doing it solo.

For myself, though, I’m more likely to accept B2’s ~1.5¢ per watch-hour rates, using it to host the actual video files (you can configure PeerTube thus), because they’ll have gigantic pipes. Then if I magically reach the point where that’s costing large sums of money, then I can optimise it.

*edit: all numbers are indeed Bit, not Bytes. Aren't all network rates quoted in bits?

In case those rates are in bytes: Streaming a 3min video at 2.5Mbitps leads to about 60Mbyte of traffic. At $1.20 per TByte, that's $0.000072 per view, or 0.0072 cents.

Or the other way round: $1.20 allows you to serve 16,667 videos.

At the moment, activists and hackers ("non-normal" people) are really the main audience for decentralized "censorship resistant" networks, since everyone else is happy with Youtube and other centralized services.

Regarding the instances page, it looks like they're pulling instances randomly, which would just be confusing to new users and in cases like you mention where the instance isn't popular, counterproductive.

The frontpage isn't much better - most of it is taken up explaining free software politics and things "normal" users won't care about, and it offers only three content examples - one video, one channel and one instance. Compare that to Youtube's frontpage, which is tiles and tiles of content, with a search bar right on top, basically choking you with content.

I understand they probably care more about the politics and philosophy but if they want to be an alternative to Youtube they need to do what Youtube does and present themselves as content first. The layout for the instances themselves do that, but the homepage needs to pull a bit more weight in that regard, since it's standing between the user and all that stuff. In my humble and unqualified opinion.

Exactly, I think the first impression should immediately show the value of the platform.

In this case it would probably be well rated videos on the default / least niche / most popular instance and not an instance with conspiracy videos or other "censored" content that can't live on the other platforms.

Being in a pretty similar situation I think the https://joinmastodon.org landing page is done in a more user friendly way. They have a "Join now" link which brings you to an instance list that has the most popular ones shown right away and even gives you a way to find out which instance is the best for your language or interests.

> activists and hackers ("non-normal" people) are really the main audience

Respectfully disagree with you. There's quite a lot of content which is getting throttled on Youtube if not outright censored.

I follow Styx on YouTube who's a libertarian and even his channel and videos are constantly censored, throttled. Dave Rubin is another one.

https://www.youtube.com/user/Styxhexenhammer666/videos

Why the downvotes? HN is getting more and more hostile.

I think it has to do with the youtube channel you shared. Best guess is that the content is alt-right or trails on conspiracy.

What's alt-right or conspiracy about it? I don't even know what the definition of "alt-right" is now a days. I am a brown immigrant myself (I hate mentioning that but it seems relevant in this context) and I find majority of the time people using "alt-right" don't know what they are talking about.

" since everyone else is happy with Youtube and other centralized services."

I also respectfully disagree, because YouTube consistently: - recommends pop-culture content that can be found elsewhere, most specifically on cable - recommends content I've already watched even though I'm logged in and they should know I've already watched it - buries the content most relative to my interests - gives me no means of adding or removing tags to help me filter my content - has search functionality that constantly breaks when one applies filters (so you do a search, get a million results, sort by upload data and it returns 'no results found' W.T.F.

Hell I'd even settle for someone to just abstract the AI/tracking/algorithm and link to content on all of these sites so I don't see content on any ONE of them and can search across all of them. Why hasn't anyone done this? If someone has, please promote it more.

Check out https://github.com/omarroth/invidious: no js needed, audio-only mode.. It has some rough edges (channel pages and a couple videos seem to break), but it's very decent.

That's why I use many accounts for YouTube, one for music, another one tech-related, ... using the multi containers extension of Firefox, this way, the recommendations become somewhat more relevant.

Ironically it’s the corporate shills and free market profiteers who end up caring about things like “what does the user want”. So many of these projects are all about ego stroking ideals and not actually providing value to anyone. There’s some food for thought in there about the parts of capitalism that work.

I think a "normal" user isn't likely to hear about PeerTube and immediately think "how can I join this?" Rather, someone might send them a link to a cool video like https://video.blender.org/videos/watch/8533ea43-4271-4a57-96... and if they like it, they look at some of the other videos on the same instance. So discovering an instance you like is not that much of a problem if you're not starting from the central project website.

Of course the instance that has the videos you like may not allow just anyone to create an account (e.g. video.blender.org doesn't) at which point it does get pretty confusing, because you need to move to some random other instance for no obvious reason. (Of course the reason is decentralization, but "normal" users don't care about that as an end in itself.)

If someone sent me that link I would be confused, since the video doesn't play and it says "0 peers" at the bottom.

Do you have JavaScript disabled maybe?

With JavaScript disabled, it doesn't do anything.

That's interesting, it definitely plays (near-instantly on clicking play) for me. Number of peers should be irrelevant, as it falls back to downloading over HTTPS. It might be worth a bug report if you can track down any reasoning.

> The first entry in the instance list is "No Censorship Tube" and you don't really see if it's an active instance, popular or anything as it just says "Follows 0 instances" which makes it seem like an abandoned project.

It looks to me as though that page presents every instance matching the user's search criteria in random order, presumably for fairness.

This is a mistake, IMO. Especially at this early stage, the project should be able to point to a curated list of a couple of well-run public instances, not a random listing of every instance they're aware of. (Especially given that some of those instances are private, empty, or broken.)

I think the killer distributed app would be some kind of peer cache of videos from yt, vimeo and so on. You watch YT and your browser stores the movie then shares it like webtorrent. G decides it doesn't like the movie, bang - deleted! Not so much, still available in distributed cache.

I look thru my yt playlist some as old as yt and half of the entries are dead. And there was a lot of original user content too. So much of human art is lost.

We have technology we need leadership!

other than the threat of copyright strikes, why should anyone allow their bandwidth to be used for this cache?

And if you want to view a deleted video, how would you go about finding it on this distributed cache? If you can search people's caches, this is the same as announcing that you have seen such and such video - which may be an invasion of privacy too.'

If the original author of the video wanted it to be unremovable, they need to put this up on their own infrastructure they control imho.

Take a moment to consider how this architecture would (or would fail to) handle truly abhorrent content -- child abuse, terrorist recruiting material, etc. Not all videos are "human art" which ought to be preserved.

Drew needs to quit spending Sourcehut's money on whatever he wants.

Back in October, he also hired "Sourcehut's first developer", but he gets paid to work on Drew's personal or favored projects (like Sway), not anything directly related to Sourcehut: https://sourcehut.org/blog/2019-10-15-whats-cooking-october-...

Sourcehut is in an alpha state, and is barely functional in a lot of ways. People are paying for it anyway because they support the site's vision and what it could be, and want Drew and the site to have the resources they need to continue working on it, so that it can eventually become something that's actually worth paying for.

Those people are paying for Sourcehut with the expectation that the money will be reinvested into Sourcehut and used to maintain and improve it. None of them are doing it because they want Drew to buy random people microphones so they can make videos for Peertube.

For someone that writes so vehemently about topics like tech and business ethics, it's extremely hypocritical for him to keep using Sourcehut's money as his personal slush fund. Maybe someday if Sourcehut is incredibly successful and they can't even figure out how to spend all their money this will be a reasonable thing to do, but right now it still needs these resources that are inexplicably being redirected to other projects instead.

The mission has always been "to improve the free and open source ecosystem", not just "to make a cool code forge". On many occasions (e.g. [0][1]) I have made it clear that the goal is to invest our profits back into the broader open source ecosystem.

[0]: https://lists.sr.ht/~sircmpwn/sr.ht-discuss/%3C2019042616072...

[1]: https://lists.sr.ht/~sircmpwn/sr.ht-discuss/%3CBVRVZEWYB30Q....

I would rather not be the kind of company which pockets millions for its founders and investors with money built on the backs of the FOSS community. An explicit, stated goal is to spread the wealth and raise up the ecosystem as a whole. If you don't believe in that, then don't pay for it, it's your money. SourceHut-the-software-forge is not going to be any worse off because we spend $5K giving PeerTube a leg up. We spend more on that for a single build host for builds.sr.ht.

Just FYI, I'm a paying customer of sr.ht and I'm glad to support what you do. You're awesome and I think this bootstrap fund is a great use of resources. Good job!

Thanks <3

First of all, bravo to Drew for trying to nudge the world in a positive direction.

I was just wondering if he was running a fund that did exactly this, to pay for a few open content creators and OSS developers. (Is there already a fund like he's sort of doing here?)

If he (or someone) did peel off these altruistic activities, they could (a) get donations from others of like mind, (b) continue to let Drew use his good sense to decide how to spread those funds out to the creators, and (c) keep Source Hut funds separate so its users got a more fair deal.

Even if he does keep them commingled, wouldn't mind paying for SH and having some of it go towards helping the community.

Yeah, I paid for SH and I'm happy to see that money going to the broader ecosystem, but I could see having a parent "advancement of FOSS" organization that then also contributed to sourcehut in addition to others. I could see an argument for sr.ht being the central entity, but it feels like an unnecessary commingling of concerns. By analogy, it feels like shipping a sha256 function in libc: It makes some sense, but I'd rather see it factored out into a crypto library.

I think this would be a good idea in the long term, and I'd like to set something like this up. Ambitions like making sure driver maintainers have access to the hardware they need, paying for security audits, and so on, would be great. The biggest resource that is scarce is time, though - so for the time being this is the easier approach.

Its users are getting a fair deal, though. They're getting the service they paid for, with significantly less downtime than GitHub, and Drew is being fully transparent about how funds are spent.

Sourcehut [is] barely functional in a lot of ways.

Since when? It has the best CI, has the best interface, has both Hg and git support, and does pretty much everything a forge is supposed to do. It also does discovery better than any other platform.

I disagree with the rest of your comment, but the rest is wrong enough and obviously in bad faith enough that a response would be unnecessary.

Yeah, just reads like someone disgruntled at how SourceHut spent the $5 they gave them.

Reminds me of the kind of comments I unlocked when I took donations for my forum and someone who gave me $10 once, a year ago, would write paragraphs about how I'm misallocating their money.

> It also does discovery better than any other platform.

So I went to https://git.sr.ht/ which shows me nothing but buttons for login and register. Same for https://sr.ht/. I'm really confused if I'm holding something wrong, but this doesn't look like discoverability at all. Is it better if I create an account?

Yes, the site makes much more sense if you create an account (without an account you can still contribute to projects, though!), but you don't have to if you just want to browse projects:

https://sr.ht/projects

This is the easiest possible interface for discovery of projects, which is (in my opinion) something that GitLab and GitHub are pretty dreadful at. I'd go as far as to say there hasn't been a forge good at it since two SourceForge redesigns ago.

Oh that's awesome. How on earth is this not the main page?

Asked; just a bug, apparently! I'll file a report later & it'll probably be fixed soon.

Thanks!

sourcehut.org is the proper landing page.

Okay, but that one does not provide discovery either. It's just a regular product landing page just like github.com or gitlab.com

Trying to sell your opinion as the opinion of a group of people is a really weak and see-through rhetorical device.

Are your roleplaying as a spokesman?

This is great, and looks similar to of Drew DeVault's ( creator of Sourcehut ) other offer of paying you to write a blog:

https://drewdevault.com/make-a-blog

I think wanting and encouraging usage of Free and Open Source software is important, and I commend Drew for putting his money where his mouth is.

If that's any indication, the previous initiative to motivate people to write a blog is a failure since most of the blogs listed there as recipients never made it past a few blog posts.

Actually, 10+ of them are still going strong. I have been meaning to sort the list by activity rather than award date. I feel like I got my money's worth with that initiative.

Can you separate them from the list?

Check back on Monday, I'll try to find time to sort it then.

Thanks, will do.

I think this is an interesting and definitely laudable project, but I think the PeerTube-exclusivity clause is a large mistep. I understand the motivations, but I think it's flawed for two reasons:

1, it increases the risk for creators. While an equipment grant is extremely useful, the biggest investment (having made a couple educational videos myself) is time. Having to then limit yourself to a platform with very low user adoption is a huge risk, and harmful to creator morale if the videos struggle for an audience due to the platform.

2, and this ties into the first problem, it's not good for marketing. If you look at how another Youtube competitor, Nebula, is handling this, they have creators cross-post most of their content to Youtube. Creators then advertise Nebula in their sponsorship timeslot, extolling its virtues for creator sustainability etc -- which PeerTube would greatly benefit from as many people aren't aware of the benefit of a distributed system, or what that even means. Additionally, they upload some sort of exclusive -- either longer videos with content that wouldn't fit on YT (LegalEagle is a good example of a channel that does this), or special long-form projects (Wendover Productions uses this model). Either way, this lets Nebula hijack Youtube's recommendation algorithm and built in audience as essentially free advertising, as the content being uploaded to YT is being posted to Nebula anyways. Obviously many people just watch on YT and ignore Nebula as it requires a subscription, but PeerTube doesn't and Nebula is succeeding despite that -- anecdotally, me and my fiance bought a subscription because of this strategy.

With that out of the way, I do love this idea and hope it gets traction regardless!

I remember visiting some PeerTube sites and being turned off by the UI. But it looks like there has been some significant improves and pageload speeds are super fast. Glad to see it making progress.

Peertube should partner with Patreon or a similar service for premium content. And hopefully more major Youtube content creators link to it in their video descriptions and other social media platforms.

Vimeo carved out their own niche with more professional content creators. Any Youtube competitor will have to find a few niches in order to seed their initial user base.

If their copyright takedown system has a less asymmetrical power balance that YT, movie reviewers and musicians might be good to target.

There is no "they". PeerTube is a software package that allows you to host your own YouTube-like site. However, it also has ActivePub (Mastodon) federation allowing you to also follow people from other instances (if allowed by your instance).

It is up to to the website's host to deal with DMCA and other complaints.

It's a little strange that they would restrict people to publish only on Peertube and nowhere else to fit the criteria. Because this makes the audience of published videos closed to zero in the current situation.

I would much rather encourage a policy of "publish on PeerTube first" as in, short term exclusivity, rather than complete exclusivity.

Forced/contractual exclusivity is bad for consumers, but short term is definitely more approachable.

From the condition:

> You can only upload videos to PeerTube - not to YouTube or anywhere else.

I'm not sure this is the best approach. As it stands now, the reality is that platforms like YouTube have vastly more users than peertube. Why not allow uploading to platforms like YouTube, but instead prefacing the video with a link to the PeerTube video? This way the content creator can get more views and PeerTube instances will get more exposure to potentially new users.

Can someone familiar with Peertube explain what the difference is between PeerTube and Lbry.tv?

https://lbry.tv/@artiintel:e/Jamaica-driving-sunset:f

It seems Peertube distributes videos using torrents or something, and lbry hosts the videos, where exactly?

See DeadSuperHero elsewhere in the thread: https://news.ycombinator.com/item?id=23200467

Honest question: how will PeerTube deal with moderation and the whole cesspit of hate problem that inflicts most user generated content sites?

It’s a hard problem that even YouTube & Facebook with all their resources struggles to manage.

Or are they planning to side step the issue, with a ‘it’s open so it’s not really our problem’ approach?

It’s up to the administrators of each instance to decide. spacepub.space will decide one thing, FramaTube will decide another, &c. Instances can decide to federate with or block other instances depending on how they like their policies and such.

Is peertube cool? I mean would the tiktok crowd like it?

My opinion: it needs 90% better UX and 10% content.

The content is a by product of having an app/site you like. Make it faster,leaner,simple and featureful. I know easy for me to say that, my point is that's what would make me use it and stick around.

Also,these platforms like peertube and mastodon really need to get on board with ditching email. I wouldn't be a HNer it HN required email (please I beg you don't change that). Same goes for reddit.

Pretend your audience is the simplest non technical person. Things like "instance" and "federated" should not be mentioned in customer facing UI at all.

> need to get on board with ditching email.

ditching email and replacing it by what? Email is federated and ubiquitous.

Nothing. Make recovery email/code/phone optional. You can't use email for 2FA recovery anyways.

I would be happy to. I looked at peertube but didn't pull the trigger and instead went to YouTube. I think the upload limits held me back. I went through a lot of sites and some would not have video descriptions, some would have no monetization methods, others no audience. I'll definitely be dropping and email, I think the 5k would be best spent adding monetization and discovery features for creators though.

Did you check out LBRY? If so what did you not like about it?

I work on https://lbry.tv and the desktop app and am just curious.

Not OP, but multiple American political conspiracy theories/theorists featured on the front page doesn't seem like a good look, or something most people would want to be associated with.

the moving thumbnails all at once everywhere is just atrocious.

That one is filed! https://github.com/lbryio/lbry-desktop/issues/3134

Just haven't gotten to it yet. We're a pretty small team right now.

Another key point is that LBRY and PeerTube have different philosophies on how to approach decentralization. I believe LBRY uses a Blockchain for media storage, but PeerTube is more built around the concept of data federation - people set up separate sites that can then seamlessly integrate with one another by passing messages around, rather than each site having to maintain a separate copy of the same ledger.

Additionally, PeerTube is also compatible with the wider fediverse through the usage of the ActivityPub protocol, meaning that channels can be followed via a dozen or so other platforms that all speak the same federation language. Theoretically, it would be possible to write an entirely new project from scratch (in a completely different language even) that's capable of federating videos with existing PeerTube installations.

This is not to say that there's anything specifically "wrong" with LBRY, just that one approach already integrates with a broader set of platforms due to what community it was developed out of.

It uses the blockchain to store pointers to find the data in a p2p network.

Anyone can build an app that interacts with the LBRY network through the lbry-sdk (https://github.com/lbryio/lbry-sdk). To be fair though, our company is the only one working on apps atm. But you could build one in any language you want.

Syncing my YT account now. I think I bounced from all the federated Peertube instances due to their upload limits. Started my account on YouPorn, they have absolutely great paid video features. No descriptions though and no audience there.

My feature wishlist would be if you can have paid videos or reasonable user access control and an API, set up some sort of shop to sell merch, and have course style organization. I use playlists for organization on YT, but it could be better. YT has a lot of that stuff, but you need to have a shit ton of subs or view hours to get access to it, so I don't have access to it.

I'll give it a shot. If you want to get in touch with me, email and Twitter in profile. I'll probably have more feedback in a month when I'm back to my recording setup and actually upload something.

Reply on news.ycombinator.com