Hacker News new | past | comments | ask | show | jobs | submit login

You’re writing this comment on a site with an upvote/downvote based algorithm.

The answer is simple, allow some level of user feedback from proven real users (for example, only people with gmail accounts that are over 5 years old and who use them at least 3 times per week to eliminate fakers—-but keep this a secret) and apply it mildly as a ranking signal.

As long as it doesn’t become the only factor in ranking, you still retain strong incentives to do all the old SEO stuff, yet with a layer of human sanity on top.




Google Maps reviews are working like that and are often gamed.

If you pay close attention, you can spot fake reviews because they usually come from “Local Guides” (so supposedly the most trusted users).

Reddit is somewhat better at ranking and filtering spam, due to local mods, like there were in the times of web directories and webrings.

One of the former bosses of Google search explained that the key metrics they follow to consider the success of “Search” are the number of page views and the total revenue.

So if a user doesn’t find what he needs but keeps coming back it’s a win for them.


> for example, only people with gmail accounts that are over 5 years old and who use them at least 3 times per week to eliminate fakers—-but keep this a secret

tbh this is just a bandage, its just going to get botted once people discover the pattern (a lot of premium bot farms do offer mature or hacked gmail accounts anyway) and its going to be worse for legitimate discovery


Then you quietly change the rule, and get years of success until it's widely gamed again.

The point is, there's experimentation that could be done and there's absolutely solutions that could be found.

Early Google did tons of experimentation with the search algorithm to maintain the integrity of the results. There was definitely an active game of cat and mouse back then that Google actually cared about staying on top of.

But as a decades entrenched monopoly, Google lost all incentives to tinker with anything anymore. The "operational" folks took over and any change to the search algorithm is now a multi-year endeavor involving thousands of stakeholders.


> Then you quietly change the rule, and get years of success until it's widely gamed again.

You won't though, unfortunately. Too many people know the rules to keep it a secret, especially given that corruption exists.




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

Search: