Hacker Newsnew | past | comments | ask | show | jobs | submit | d10r's commentslogin

Can't confirm, have been running stuff on Hetzner for almost 10 years. They will indeed auto-block you if any application is spamming the private subnet - happens e.g. with IPFS running in its default config. But this always comes after warnings and with an exact explanation of what happened (log of IPs the server attempted to connect to, timestamps etc). I always found support responsive, even in the middle of the night. It may be that I was lucky all the time, despite interacting with various support staff in various of their data centers. Or it may be that your issues had other reasons...


Bitcoin would have remained the obscure crypto-economic experiment it was 10 years ago if it weren't for the rotten monetary system we have creating strong incentives to look for alternatives. I'm holding some Bitcoin for quite a while now - in terms of value it has overtaken all my Fiat holdings - yet I would actually welcome if it became worthless. That's because I believe that it can become worthless only if we get back to a sound monetary system, something I'd value more than individually "winning" in an alternative scenario.


which may even be technically true - if measured in quantitative terms :-)


As long as we stick to the concept of nation, we likely have to live with such issues. This is how group identity works, sometimes to your advantage, sometimes to your disadvantage.


Here in Europe it's often the opposite. E.g. in Austrian cities you have decent bicycle infrastructure and everything in reach in the cities while in rural areas that's often not the case.

I've been in the US once, 13 years ago, and it was pretty shocking for me to experience the concept of "car centric" in its full glory for the first time. I was at CES in Las Vegas and went to some club one evening with a friend. At some point I left and wanted to walk to the Hotel alone in order to calm down and enjoy the nice climate. Turned out, there was simply no walkable connection between the 2 locations. I couldn't believe it, but - being stubborn - walked anyway, in the dirt along some highway, a bit scared of being picked up by the police, not even sure if walking there was even legal.

Later that week I moved to LA and first saw the endless suburbs of an American city, from the air.

I don't know how representative those 2 places are for the US, but having seen that, I can totally understand why many Americans have a very hard time imagining life without a car.


Unfortunately it's not as simple as that when large scale network effects are involved.


You should probably not listen to random YouTubers, but to people having something to say. I suggest starting from aantonop.com


That's indeed true for many DAO designs. I often wonder if those proposing such designs don't understand that or if they just don't care. It is however not inherent to DAOs. If you add identity, a lot of alternatives open up. Check out e.g. quadratic voting or daostack.io.


> Could there be a class action lawsuit against the various companies whose recommendation engines hijacked people's attention to recommend and reinforce this garbage?

I think this thought is spot on. The usual defense is "but free speech!". Which would boil down to: "such is human nature". But I don't believe that's the problem. The problem may indeed be selection and amplification mechanisms like recommendation engines tuned to divert max. attention to the medium, masterfully exploiting the vulnerabilities of the human psyche as evolution formed it. The rest is collateral damage which nobody seems to feel responsible for. Not a sustainable situation.


A good legal question is whether the selection/rejection of content by an algorithm tuned to provide financial benefit to the platform would be considered "editorial control".

If there were a dead simple filter where the user could pick friend groups, tags and sorting criteria to tune their feed this would not be an issue. Reddit, for instance seems relatively simple in that respect- the presentation is a function of the subreddit and the votes.

But once the algorithm is driven by sponsorships, monetization opportunities, and opaque surveillance data then the control of the presentation shifts from user to platform. One could argue that this should creates some liability on the part of the platform.

If the editor of a publication was monetarily compensated for publishing lucrative slander, in a manner designed to maximize its credibility with certain audiences, and it resulted in harm to people then they arguable could be held responsible. If the editor claimed that an algorithm decided to publish it and the publication developed the algorithm I don't think it would make them less responsible.


you're omitting something very important in your list of related aspects: health


Guidelines | FAQ | Lists | API | Security | Legal | Apply to YC | Contact

Search: