Small Phone Farms vs Big Farms The Result Usually Surprises People

Davidwolf

Regular Member
Jr. VIP
Joined
Dec 28, 2024
Messages
396
Reaction score
367
Something I’ve noticed after running different phone setups for YouTube over the years.


Most people entering this space assume the advantage is scale.
Build the biggest farm possible.


More phones.
More accounts.
More views.


Sounds logical.


But in practice I’ve seen smaller farms outperform bigger ones more times than people expect.


Smaller setups naturally move slower. Watch sessions spread out, traffic builds gradually, and the activity pattern ends up looking much closer to normal viewer behavior.


Large farms introduce a different kind of problem.


Once dozens or hundreds of devices start pushing activity under the same logic, patterns begin to show up. Sessions start overlapping, watch behavior repeats, timing becomes predictable.


From the outside that traffic stops looking like independent viewers and starts looking like a coordinated system.


That’s usually when things start getting unstable.


Ironically some of the YouTube setups that lasted the longest weren’t the biggest ones… they were the ones that grew slowly and stayed a little messy.


People chasing scale often forget that on YouTube predictable behavior is usually the real footprint.


Anyone here running phone farms for YouTube for a while has probably seen something similar happen?
 
This hits close to home, and I think the principle applies across platforms.
Yeah @TheReviwer I think you’re right about that.


Once someone spends enough time running these setups the pattern starts showing up almost everywhere. Different platforms, same underlying issue — scale tends to create patterns faster than people expect.


What I’ve seen with YouTube specifically is that the farms that stay alive the longest usually aren’t the biggest ones. They’re the ones where the operator had the patience to let things grow slowly.


A lot of setups actually break at the exact moment they start working… because that’s when people try to scale too fast.


Suddenly the same logic is running across dozens of devices, sessions start lining up, behavior repeats, and what used to look like scattered viewers starts looking coordinated.


That’s usually when instability begins.
 
Do you think the solution is just spreading the activity more randomly between devices, or is it more about separating groups of phones so they behave like completely different audiences?
I think randomizing timing helps a bit, but that alone usually doesn’t fix the real issue.


A lot of people focus on spreading activity across devices. Different hours, different intervals, etc. But if all the phones are still behaving like the same type of viewer, the pattern eventually shows up anyway.


Something I noticed after running farms for a while is that YouTube seems to react more to audience behavior than just device activity.


For example imagine an apartment building. Maybe 20 people there watch the same video. Totally normal. But those 20 people also watch different channels, different topics, disappear for a few days, come back later.


Where farms start looking weird is when hundreds of phones slowly turn into the same viewer. Same watch path, same channels, same rhythm.


At that point it doesn’t really matter how random the timing is.


From the outside it starts looking coordinated.
 
bro why the fuck there is so much ai bots in here. bunch of gpts speaking with each other with no real info.
bro why the fuck there is so much ai bots in here. bunch of gpts speaking with each other with no real info.
honestly I wish I knew too.


The funny part is I’ve realized recently that I’m actually getting bad at telling who’s a bot and who isn’t anymore.


In many of my threads I’ve been reading replies thinking “ok this one is probably AI”… and then the person comes back later with a normal comment and I start doubting myself.


It gets even worse in my sales thread. Sometimes I see a question that looks like a bot question so I just ignore it. Then a few days later the guy shows up on WhatsApp asking why I never answered… and it turns out it was a real person the whole time.


Happened more than once already.


At this point I’m starting to think the real danger isn’t just bots posting… it’s that we’re all slowly losing the ability to tell the difference
 
honestly I wish I knew too.


The funny part is I’ve realized recently that I’m actually getting bad at telling who’s a bot and who isn’t anymore.


In many of my threads I’ve been reading replies thinking “ok this one is probably AI”… and then the person comes back later with a normal comment and I start doubting myself.


It gets even worse in my sales thread. Sometimes I see a question that looks like a bot question so I just ignore it. Then a few days later the guy shows up on WhatsApp asking why I never answered… and it turns out it was a real person the whole time.


Happened more than once already.


At this point I’m starting to think the real danger isn’t just bots posting… it’s that we’re all slowly losing the ability to tell the difference
Exactly.

Honestly, I believe the only way for BHW to survive as a service on the horizon of 3-5 years is to make registration paid. Because these bots are a pandemic.

With paid registration creating bots would become not rentable.

I didn’t post this in BHW suggestions forum though, because nobody will support this idea anyway.
 
I’m with you on that @NomixGuy


Feels like fully free access is what opened the floodgates in the first place. I don’t even think they’d need to lock everything behind a paywall… but putting some kind of friction on posting would already clean up a lot.


Something like letting people browse and maybe interact lightly for free, but requiring a small fee to reply or start threads would probably kill most of the low-effort bot spam overnight. Right now it’s just too cheap to abuse.


I’ve also noticed what you’re talking about with lifespan. A lot of these accounts don’t even last long, sometimes 24h and they’re gone. Mods are cleaning, but the damage is already done in terms of noise.


At this point it’s less about stopping bots completely and more about making it expensive enough that it’s not worth scaling.
 
Something I’ve noticed after running different phone setups for YouTube over the years.


Most people entering this space assume the advantage is scale.
Build the biggest farm possible.


More phones.
More accounts.
More views.


Sounds logical.


But in practice I’ve seen smaller farms outperform bigger ones more times than people expect.


Smaller setups naturally move slower. Watch sessions spread out, traffic builds gradually, and the activity pattern ends up looking much closer to normal viewer behavior.


Large farms introduce a different kind of problem.


Once dozens or hundreds of devices start pushing activity under the same logic, patterns begin to show up. Sessions start overlapping, watch behavior repeats, timing becomes predictable.


From the outside that traffic stops looking like independent viewers and starts looking like a coordinated system.


That’s usually when things start getting unstable.


Ironically some of the YouTube setups that lasted the longest weren’t the biggest ones… they were the ones that grew slowly and stayed a little messy.


People chasing scale often forget that on YouTube predictable behavior is usually the real footprint.


Anyone here running phone farms for YouTube for a while has probably seen something similar happen?
Ahem, nope, this is just plan logically and technically wrong and a subjective opinion, if every device is set tup properly , the smaller farm has not a chance in the world to keep up, do not overthing it lol
 
Yeah let's use AI to detect AI... honestly, not impressed by this idea.

In the end, no matter how good moderation is. If your community is overwhelmed by cheap bots, it will just be easier to fight the system than accept the rules. That's why I only believe in paywall. But like I said, I don't think owners will do it, because it's too risky to loose audience. But at the end quality of posts will drop down significantly enough to loose audience anyway.
 
Yeah let's use AI to detect AI... honestly, not impressed by this idea.

In the end, no matter how good moderation is. If your community is overwhelmed by cheap bots, it will just be easier to fight the system than accept the rules. That's why I only believe in paywall. But like I said, I don't think owners will do it, because it's too risky to loose audience. But at the end quality of posts will drop down significantly enough to loose audience anyway.
You see an iceberg.

You only see 10% because the other 90 is under water.
 
Ahem, nope, this is just plan logically and technically wrong and a subjective opinion, if every device is set tup properly , the smaller farm has not a chance in the world to keep up, do not overthing it lol
i think there was a bit of a mismatch in what i was trying to say.

i’ve also run volume-focused setups before, so i get where that “more properly set up devices always win” logic comes from. inside the typical panel model, that’s actually true.

but the point i was making isn’t about delivery capacity, it’s about signal quality.

once you start looking at it that way, scaling isn’t just about adding more devices. the bigger the farm gets, the harder it is to maintain consistency in behavior, human patterns, natural distribution… and that’s where the gains stop scaling the way people expect.

a “perfect setup” on paper is one thing, keeping that running clean at scale is a very different game.

that’s why i said the result usually surprises people. it’s not that big farms don’t work, it’s that they don’t always perform better in the way most expect today.

in the end it really comes down to what you’re optimizing for: raw volume or how the algorithm interprets those signals.

are you mainly running volume right now or have you tested more controlled setups focused on that side?
 
This matches what I've seen - bigger farms fail not because scale is wrong but because scale amplifies whatever's already detectable. 20 devices with identical behavior/IP structure is 20x the correlation surface. Small farms surprise people because the footprint is small enough to stay under clustering thresholds. The operators who scale successfully are the ones who make every added device look independent (own fingerprint, own network, varied behavior). Scale is a multiplier on your hygiene, good or bad.
 
Back
Top