Show HN: ProveTrust, a user trust reputation system for the web | Hacker News Reader

Show HN: ProveTrust, a user trust reputation system for the web

provetrust.com

I looked into jumping into this space recently, and doing my research it seems that all the companies doing this exact same thing (and there are lots of them) make the same mistake that this one does: Involve the customer in the loop.

This requires Joe AirBnB User to actually sign up for an account with the service (and a dozen others) and tell it about all the places that it should agglomerate reputation from. That's precisely backward.

This service's customer should be AirBnB, eBay, etc. And if the big ones don't immediately sign up, you'll need to go convince (as in, pay) them to partner up from the start. Customers get matched up with profiles behind the scenes automatically and end up with a "credit score" assigned to them that they have no control over.

The company that builds that product will win. The other twenty will die, as they are currently in the process of dying, because nobody is going to wake up in the morning, Google "centralized trust service", sign up, and spend an hour filling in information.

We tried this and failed. None of the big guys want to share any data with you. They want to build their own reputation system.

Have a look at these guys. They found a nifty way of getting people to sign up once.

https://www.trustpilot.com/

But yeah the problem is still to get people to me active providers rather than just ad-hoc.

Very good points about user friction, which should be minimized. The plan is to make adding information from you accounts MUCH easier. Stay tuned.

This. Also exactly why each ecosystem has their own trust mechanisms built right in.

Trust based on social media: -1

I remember Thawte used to have a WOT (web of trust), and "web notaries." If was an interesting idea, based on personally meeting, providing passport as a proof, etc. and have people already trusted vouch for you. They shut it down.

Thanks for the feedback. I plan on adding other forms of login in the future; Facebook just has the largest user base.

I'm with the parent commenter. I actually deleted my Facebook, but would love to utilize this service to create a public trust profile. I just can't involve social media accounts.

What other accounts would you be comfortable associating with this trust profile? Would you hook up your GitHub?

I'd hook up my PGP key. Or X.509 certificate. Or old good username and password pair. Something I can truly own and handle. Why do I need some third party to have an identity?

Then I'd enlist accounts with any imaginable third parties (email addresses, Facebook, VKontakte, Google+, BitCoin addresses, GitHub, BitBucket, StackOverflow and so on and so on) as something I have at the moment.

The idea is to have a central place on the web to store user reputation and reviews across many sites. You should be able to have one reputation (Trust Score) for all sites and not have to re-build it for every eBay or AirBnB. Further, a lot of sites like Craigslist or OKCupid don’t even have reputation systems, but could benefit from one.

I consider this a minimal viable product and plan on adding a lot of features, but would like to get some early feedback. Thanks in advance.

I worked for a company that would have been a target client for a product like this. In the end, we didn't use one of the many similar solutions (legit.co, trustcloud, scaffold, peertrust, project trust, truly, etc..) for two reasons:

* We didn't want to give our information away. It was a nonstarter. Either to protect it from being used by competitors or because of security concerns of giving private user information to a third party. * Each site and community has different levels of trust, and actions have different meanings. Whether someone commits sexual harassment on OkCupid doesn't matter when you are trying to buy their toaster on eBay. I am much more concerned about someones behavior when they are living in my house, not so much when they are giving me a ride in their car within the city at noon.

It would be nice if everyone didn't have to build their own trust system, but any competent company will. A general solution will either be too general to be useful, or require too much private information to be allowable.

The only way I can see a standalone product like this occurring is if a well established company like AirBnB of Facebook would build it on their existing (valuable) dataset and and guarantee data privacy.

I worked for this same company, and wholeheartedly agree. This type of product is a non-starter. Companies like Legit.co, Trustcloud, and Scaffold have all tried this and failed.

Furthermore, there's really not much value in the scale you're giving. It's scary. If someone is 95% trustworthy, should I trust them? Or is that last 5% the axe murderer portion?

Overall, I can't see any target market for this. Companies would build their own, and consumers don't want to take complicated 3rd party trust into their own hands.

Thanks for the feedback. In this case, the target clients are regular users, not companies. Craigslist doesn't need to share any of their data for this to be useful to Craigslist users.

"Whether someone commits sexual harassment on OkCupid doesn't matter when you are trying to buy their toaster on eBay."

You can choose to ignore this information, but for some users it may be a relevant signal. I personally would rather buy a toaster off of non-sexual-harassers. Also, perhaps there will be less harassment if there were consequences, i.e. a worse online reputation.

If data isn't coming from service providers, does that mean it's all coming from user reviews? ISTM that e.g. Yelp (there are others too) has been working through the inherent difficulties for some time, and they haven't exactly got it figured out yet. What if one of the first million reviews/ratings leads to a lawsuit? Does this all just go away?

Looks very polished for a MVP, well done!

I do not have a Facebook profile, so the potential employers that run an "internet background check" will not trust me much.

Does that mean that I am not trustworthy? What I am getting at is that there are multiple levels of real and artificial trust. Being socially engaged on Facebook shows just that: you are are socially engaged and have lots of friends: you are probably a real person. Scammers can be very sociable people too, yet only their closest friends can hope to trust them, for the rest it's a facade: keeping up appearances. Buying followers to increase their Klout score.

Let's say this gets big. Do you expect attacks on your system and algo's to inflate the trust ranking? Would people leave fake reviews for themselves? Would people link up fake or stolen craigslist profiles?

If you happen to get the snowball rolling and the demand gets very high, do you think companies will form to cater to that demand? Scammers could create profiles that score well, and sell them priced according to their Trust Score. A luxury problem to have, but what is your view on this?

Finally how do you deal with people sending you e-mail or letters complaining their score is too low, or that your algo's are unfair, or that they didn't create that public profile, or want to remove a negative review? There seems quite a lot of (legal) hassle involved in the user problems that arise with such an application.

Thanks for the feedback. I plan on adding other ways of signing-up to serve the folks without an FB account.

Even if you don't have a lot of friends, you can still collect feedback and add badges, which will increase your score. The social aspect is just one component.

Regarding fake reviews, I have a background in Spam and Abuse. I look forward to tackling the fake reviews/users. Unhelpful reviews will be ignored and removed.

Regarding the algorithm, that will be continuously tweaked based on feedback.

Someone who is a trustworthy seller may not be a trustworthy date partner, do you generate different scores based on context or just average them all?

Good point. The scoring is going to be continuously tweaked as more data is collected. It may be the case that your dating behavior is irrelevant to your Craigslist behavior, or it could be a useful signal. The scoring will adjust according to the data.

Another thing to consider is long term versus short term behavior. In my experience, some people have spurts of negative activity that are important to account for in small time spans but which may smooth out over time, that is, you may want to steer clear of someone for a month but the next month they are fine. This can coincide with financial issues, mental health as well as general stress.

great job so far. what stack you using?

Thanks. Python App Engine with AngularJS.

Very cool, have seen a few sites doing this sort of thing. The only concern with these type of services is the fact that scammers will harvest trust on small trades and then scam big trades.

Your service should verify the real life identities of users or perhaps work in some sort of escrow.

Best of luck

Thanks for the suggestion. Yes, this is a common tactic that scammers on eBay do. I will work on mitigating this form of abuse.

Any way to tie into Keybase (https://keybase.io/)? They already have my public key and several identities.

An interesting direction would be to build an open Rapportive. Index and associate everyone you can think of and make the information searchable. E.g, John Doe == facebook.com/johndoe == linkedin.com/profiles/john.doe == twitter.com/johndoe == www.jdoe.com

Here is my trust profile. https://www.provetrust.com/#!/p/HJoWjhUC7JW8u9eiBVMh7E

So the deployment model will be like Disqus, you just add some JS to your page?

I wanted to do something similar but was worried about users putting negative stuff on other people's reputation as a way to harass them.

Reply on news.ycombinator.com