Hacker News Reader

Comment by toasted | Hacker News Reader

It is only a matter of time before these outfits are replaced by decentralised, actual peer-to-peer networks, either running off blockchain technology or similar decentralised databases, with no house take and the only option for government being charges against individual operators. There is no network lockin effect when drivers can easily install a competing app and drive for no commission to anyone else.

View on HN

...and nothing stopung Sybil attacks from forging quality records/reviews/feedback, or bad actors (criminals, kidnappers, rapists, et c) from participating.

No thanks. Did you ever stop for a moment and think that perhaps people use Uber for a reason beyond "they developed an app"? It's the standard of service, silly.

> and nothing stopung Sybil attacks from...

Nothing except that fact that you and every other customer can choose to not patronize any drivers or services that don't provide the information or assurance that you require.

Yes, that strategy will work fine for my wife when she's leaving work at 3am and needs to get back home. It's fine, she can walk the seven miles - her knee doesn't begin to hurt until about five miles, anyway.

She could take whatever taxi or ridesharing service she thinks is legitimate. I doubt your wife is so unusual that there isn't a market for the type of car service she desires.

Theoretically, couldn't a decentralized ride-sharing system be made sybil attack-proof (or at least robust to it) with a feedback mechanism built on transitive trust?

That sounds awesome- you'd get all the safety and accountability of hitchhiking, plus you'd have to pay for it!

A properly implemented reputation system works great. In addition to this the network could implement certification systems which allow trusted authorities to issue qualifications on certain drivers, eg. for those with a clean criminal record or those who have passed a local area knowledge examination

Out of curiosity can you point to a "properly implemented reputation system" that "works great?" And please don't cite online drug or carding markets they have centralized vetting, escrow services, high barriers to entry on both sides, and HUGE legal incentives to play nice and avoid legal attention.

How about the vast majority of the restaurant industry?

The basic business model of a restaurant is repeat food sales. But some criminals are opportunists who are not worried about repeatability; they heavily discount the future in the face of a sufficiently attractive short-term payoff.

That's true, but the majority of taxi/ridesharing drivers, and the vast majority of honest well-behaved drivers, will want to be in the business for an extended period of time. There are various methods to pre-filter potential business partners by estimating their honesty. Reputation systems and signaling are two broad methods which I think can be adapted to the ridesharing sector.

I do not share your optimisim about reputation systems. One of the benefits of an employer (who can be regulated) is that they have an economic incentive to do that filtering on behalf of the customer in order to protect their own bottom line. I don't want to spend time figuring out if a cab driver is legit or not, life's too damn short for that. That's part of what I am paying for when I hire a cab.

Meanwhile Uber has another PR headache as of last night... http://www.sfgate.com/crime/article/Uber-driver-accused-of-h...

I disagree. People were saying the same thing about Facebook and Diaspora but that never actually happened.

That's because diaspora sucked. The distributed social network is coming.

Wow, this is the impatience of our age. How long do you give a network distribution methodology before you declare it a failure?

Facebook has a strong network effect in that everyone is on it and few regular people are interested in signing up for multiple social networks.

OTOH, rideshare drivers are incentivised to have as many (useful) rideshare apps on their phones as possible to maximise their ride volume, and customers are then incentivised to order their rides from the rideshare app with the lowest commission, and eventually a race to the bottom will result.

I think more people are willing to switch away from Uber/Lyft than from Facebook since (from what I understand about the service) there's nothing really personal about them except for user ratings. If a new competitor opened up tomorrow that promised to be decentralized with fewer fees, I'd bet on it succeeding. Facebook has many more purposes like communication, media hosting, and event planning (with a very strong network effect). People aren't going to give up on that platform very easily.

I prefer having a third party to ensure quality and have recourse for disputes (it's nice to know that you can sue a company with $1,500,000,000 in venture capital if something goes really awry).

That said, drivers for these services are going to be replaced by self-driving cars.

A robust reputation system using chain of trust scoring to reduce the usefulness of sybil attacks would rapidly result in reliable, friendly, safe drivers building huge amounts of positive feedback, which would then self-reinforce improved customer service efforts to maintain their reputations.

If the online drug markets can use reputation systems to ensure fairplay amongst the vast majority of anonymous blackmarket trades then local taxi networks should be able to manage it too!

If you get ripped off on an online drug market you don't have a recourse (a poor feedback, but hey then you'll get a poor feedback back).

Similarly, I like that I can call my credit card company and dispute a charge. The blockchain style of trust just means you get screwed.

Are online drug markets decentralised?

they increasingly have decentralised aspects (eg. multisig escrow) and the new openbazaar is truly decentralised

This is wrong on numerous levels. For starters, such a service would be materially inferior to what Uber is providing. The payment mechanism is almost irrelevant. What you describe would not be any more legal nor much harder to police. And you drastically overstate the commodity nature of the service (in fact, it's a winner take most/all scenario).

Reply on news.ycombinator.com