Let’s talk about spam

Some of you have shared your frustrations lately about sketchy upvotes, spammers soliciting their services, and bot or AI-generated comments on the site. We hear you. Launch day is a big day for all makers, and we want Product Hunt to be a level playing field for all makers to get support and feedback from real users (and have a fair chance at success).

This isn't a new problem for us or any community-driven site that reaches scale. Within the first few months of Product Hunt's launch, we saw fake accounts appearing on the site and upvoting products in an attempt to game the rankings. At the time, we had very little in place to combat this. Our moderation tooling was Ryan Hoover, waking up at 5 am to delete obviously fraudulent upvotes. Literally.

We have a few more people and tools in place today ;)

The big challenge in managing moderation and spam is that it's opaque by design. Over the years, we’ve added new tools and methods to combat inauthentic behavior. At the same time, we cannot expose some of the techniques we use to detect and remove bad behavior as doing so makes it easier for shady folks to exploit the system.

Another challenge is managing perceptions. Some enterprising people promise to guarantee the top spot on Product Hunt. Unfortunately we have no control over the emails or DMs they deliver off of the Product Hunt platform, and we know that many of their claims are untrue, but their mere existence can cast doubts on the integrity of the ranks. Fortunately, we're very familiar with these actors and their tactics. We've unlisted products that use these "services" and will continue to do so.

To be more upfront about the other steps we’re taking, we want to share a few new developments we’ve been working on behind the scenes.

We have and will continue to adjust the homepage algorithm to quarantine and disqualify inauthentic or bot-like voting behavior. Recent work by the engineering team has helped us create better moderation tools and programmatic detection, dramatically improving spam detection. We’ve also been significantly ramping up the aggressiveness of our policies in the last few weeks; we know that some people are buying votes, our detection of this is *very* good, and our moderation actions will continue to get increasingly severe as necessary to ensure cheaters never prosper. Don’t be that maker whose (otherwise great!) product gets removed from the homepage due to shady tactics.

We’re testing a randomized homepage sort at the start of the day. This gives all products on the homepage a more distributed chance at exposure early on their launch day – something we’ve gotten requests for time and time again. Starting tomorrow, randomization will occur for the first 2 hours (PT), but like any experiment, we will likely make adjustments based on results and community feedback.

Update: We have upped the amount of hours the homepage is randomized for to the first 4 hours (PT).

We’ve released a few features recently with a focus on improving user trust. You can find these in the new ‘Trust and verification’ section in your profile settings. This includes LinkedIn login as a way to verify your Product Hunt profile (which also appears on reviews that you leave across the site), a more significant focus on email verification, and surfacing the age of accounts. We’ll continue to improve and develop this over time.

The ‘report’ link has been around for a long time, but it has become more important in recent weeks as we’ve seen an increase in ChatGPT-created comments. We had a great discussion with the community about AI comments, and although automatic detection comes with a whole host of challenges, it’s definitely top of mind. Last week, we added an AI-specific flag to the ‘report’ link as well, so please help us by flagging any bot-generated comments you find around the site.

We have also introduced features to encourage even more thoughtful community engagement, like Comment Prompts and Comment Awards.

We have lots of automated tooling, but sometimes there’s no replacement for real live humans. We have doubled our human moderation team to ensure bots are flagged quickly and don’t impact the homepage rankings.

Like any social platform online, spam is a never-ending battle, but we’re committed to continuing to invest in it, so Product Hunt remains the special place on the internet that we all know and love.

We appreciate your patience as well as your feedback, so continue to let us know how we’re doing and if you have any suggestions by commenting below or requesting time with our team through Community Hours.

Thanks,

Ash