AMA - is there anything about how the Warpcast feed works that you'd like to learn more about?

Technowatermelon. Elder Millenial. Building Farcaster. 

nf.td/varun

/new-york by day 🗽 /latenightcrew by night :) Fan of eating tacos and bringing people together, often at the same time 🌮

If there is a reply from the author or someone you follow, its shown in the home feed. 

If there are multiple options, we choose the author first and then the one with the most engagement. 

Yes, we want to do multi-level replies, just haven't gotten around to it yet.

How does reply bumping work? How do you decide which replies get shown in the feed vs. not? Do you ever show replies below a depth of one and if not have you considered it?

Exploring what's next. Founder /launchcaster. YC alum. Avid trail runner 🍊🏃

Host /health-hive
| /physicaltherapy | Anime | Fitness 

The primary factor is velocity of engagement - so if you get a few likes in the first few minutes its much more likely to make it up there than if you get a few likes over 20 or 30 minutes. 

We also filter out spammy engagement to make it harder to game this.

I know I’ve asked one

But one more question 

What is the prerequisite for making it into the “trending casts” section

Sometimes i see casts made a minute ago in the “trending “ section and wonder

How are they trending already? they just got posted into the feed.

Building Higher apparel at High Line 🧢 
Host at /clickcast /cpg

using embeddings to find similar content to recommend.

What is something you guys haven’t tried testing yet that’s on your radar still?

the golden rule is - don't reply to people who don't find your comments helpful or posts in channels where the mods don't find your content interesting. 

a single bad post won't do anything, but if you keep repeating the behavior you'll get flagged as spammy.

I want to know about the spam rules and how to get out of it

Thank you

No, you can see casts from people you don't follow if: 

- they're from a channel you follow 
- they're liked by a lot of people you follow
- they're recast by a few people you follow

Is it only through the boosts distributed to users daily that our casts will be able to be discovered by other users now and in the future?

Stay Curious // iixii // Lazer Technologies // Prev: CoinDesk, Roc Nation // based

in the feed? 

we only show the top level replies today, plan to add functionality to show deeper threads in the future.

Why do comment threads get hidden? Even when the most liked cast in the back and forth is farther down the reply chain?

regenerative gambler @lottopgf by /ottodao | playing /lotto and uniting /degensxregens | Member @safary 🦁

you'll see recasts if at least two people you follow recast something. 

it used to be one, but this led to a lot of spam breaking through so we raised the threshold.

what happens to regular recasts? where/when do they show up?

Follow to get the best of Farcaster using @mbd - AI Recommendation Engine for Web3. @a16zcrypto css23. Ex-ML Lead @Chainlink. Ex-Sr Data Scientist @AWS.

we've mapped out the most active accounts to specific topics like "ethereum" or "art" 

if you sign up and choose ethereum as your interest, we'll show you up to 20 posts from each account that's the interest category so you get a chance to like or follow them.

Do you use any exploration mechanism for new posts, i.e how is the cold start problem solved?

Creating brand magic with startups @byronfive | Ops expert 👧🏻 👧🏻 👦🏼 | NYC or nowhere🗽| Prev Co-founder COO/CMO @snickerdoodle, Citi, Amex | /hopecore 🤞

each channel owner can add moderators or tools like autobot and set up their own rules for what makes it into main.

I don’t see a dramatic difference between main and recent feeds. What is the benefit? What determines what goes where?

we have some heuristics which look for: 
- accounts that are active (posted recently) 
- accounts that haven't been reported as spammy 

we then order by engagement they've received recently and mutual follows and then rotate through them so you don't see the same ones too often.

Feed adjacent: how are suggested followers determined? 

Could there be a section for newly active accounts to help surface new people to follow and to encourage the existing user base to welcome and engage with new accounts?

in the home feed, you'll see up to one reply from someone you follow. 

if you tap into the cast, you should see all replies from people you follow.

How can I see comments of people I follow? What if those comments/replys generate a lot of interest? Will they ever be visible in the feed?

cooking @askgina.eth | prev: blockchain research @coinbase | sidshekhar.com | askgina.ai

yes, interests map to a set of accounts. 

we show you up to 20 casts from each account that is linked to your interests so that you have a chance to follow them.

How are the initial feeds constructed for new users now (without the auto follow list). Is it completely based on the interests they select during onboarding to warpcast?

Founder https://tuum.tech | @intori | /intori | DonaldBullers.twitter

they aren't prioritized any differently than other kinds of casts.

Frames. How are they prioritized in the feed? Future plans for showing collapsed vs expanded? Seemed to be a step function innovation that needs a little more business use case clarity

ENS collector and trader.  Every domain is special, and I like to tell their stories. https://vision.io/profile/0x08e17C7912E48ac5EC61399a05A362aD64963db8

we don't yet understand semantics - it's on the roadmap. 

but if you like casts from an author or channel you follow, they'll show up higher in your feed.

Do you factor in my likes/recasts, when you try to determine what to include in my feed? If I keep liking casts about a specific topic, will this come up more often in my feed?

Art enthusiast, passionate about mental growth, and an ever-learning individual

it's engagement from people over time, minus engagement from spammy users. 

root casts only for trending.

This is probably a popular one, what’s the requirement to get your cast into the trending casts

Could comments also make it into trending casts, or does it have to be root cast only ?

the intern / the best place to discover, buy & trade your favorite digital collectibles (yes, that magic eden)

Content not intended for US/UK audiences

follow more people or like stuff from people you want to see more casts from

i keep seeing casts on my feed from the same set of people. is there a way to fix that?

Building and publishing Farcaster growth playbooks. Chief builder @percs

we run lots of tests and we use StatSig to run them as A/B tests

How do you determine which tests to run and how do you measure the success of these experiments?