Do you speak DAU ?
What are DAUs ?
DAUs stands for Daily Active Users. This term is often used in the context of web analytics, social networking services, online games and mobile apps to indicate the number of unique individuals who interact with a product or service within a single day. It's a key performance indicator (KPI) that helps companies measure user engagement and growth.
Same as DAUs, you have WAUs, (weekly), MAUs (monthly), etc.
If you want to read more, the Wikipedia article about Active users is massive !
What are they used for ?
Like any metrics, they are used to measure performance or trends. They are a common KPI to track growth of a social media or an app.
They are part on the non-financial metrics issued in reports of listed companies.
Are they good or bad ?
There is no good or bad metric. There is only good or bad usage of numbers, or misinterpretation. It is commonly admited that DAU and MAU are good metrics to measure the growth of a network. It can be biaised if users are not unique individuals and that's what you value, or if you have an unclear definition of what is active.
At the moment, it appears to be trendy to consider DAUs as a "fake" metric to measure Farcaster growth, as people state that "there is a lot of bots".
- Bots pay 5$ to FC to sign up.
- Bots circle-jerk to amass DEGEN tips.
- Degen tokens cover the signup fees.
Rinse and repeat.
FC adds feature to hide bots. "Win-win". Users live in a bubble, FC can "demo" insane DAU. Completely healthy ecosystem /s
192
Humans vs robots
Did you say bot ?
Dan says it all. There is no exact science when defining what is a bot, or a bot-like behavior.
Some accounts can have a bot-like behavior all the time, or alternate between programatic and manual casting. There is no exact science.
Hunting bad bots looks like a game that's going to be entertaining FC for a while.
Let's see how we can craft some weapons.
Chasing "bad" bots
In a world where content is AI generated (not 100%, hopefully you'll find some typo in this article), let's delve look into how we could identify a "bad" bot.
Bad bots are bad because:
They create content that does not bring value. That's subjective. ❌
They have a robot-like behavior: mechanical, emotionless. Like most humans when the tip $DEGENs. ❌
They are low value accounts, without any verified address. That's a problem for new comers and non-crypto-native users, but it can be a good insight in relation with the previous criteria. 😐
They are super annoying to real humans. How do we identify real humans ? They have doxxed previously and have verified addresses. ✅
It's easy, all we have to do is to ask real humans to filter other annoying users...
That's a lot of power to give to people, and might end up badly. Also, not very scalable.
But wait, what if we just prevented bots from being bots, ie casting annoying stuff.
I will write a dedicated article on how signers work, and to understand the following proposal you have to get it. If you are scared of slightly technical topics, you may want to jump to the next section.
No key, no cast
In order to create a programatic bot, you need around $3 worth of ETH on Optimism to create your FID and get storage, some dev skills, and an autogenerated signer or a signer generated from an app that can be used to casting programatically.
Without a signer, you can't cast. A signer is always attached to a signer app, and signer apps are identified by their FIDs.
Anyone can create / be a signer app, hence anyone can create signers, hence anyone can give "bad" bots the keys to casting.
What if a DAO was responsible for controlling signers creation and usage, in order to limit the temptation to create bot apps and monetise them ?
The bot-or-not frame already showed a PoC on how users could collectively assess bots.
Plug that with the information about the signer issuing the cast, and you have a way to warn signers issuers that some of the content casted using signers they have issued is spammy / bad ?
In most cases, that might even be enough for the signer issuer to warn their user that they have to stop their spammy behavior.
If the situation does not get better, the DAO could use a way to flag the signer issuer as promoting bad content, and the same way SMTP servers get blacklisted, signers issued from these signer apps could be blacklisted for consumer clients.
Pros:
The Farcaster network stays decentralised and permission-less
It makes it harder for spam bots to get a good ROI
The great power of signers issuers is balanced with some responsibility
Instead of fighting the bots one by one, this should take down bots in bulk.
Cons:
The blacklist, even if controlled by a DAO and community-fed, is somewhat centralised
The signers issuers have a lot of responsibility and risk, which is bad for builders.
What if bots use browser automation with consumer apps ?
In that case, the same solution can be applied.
The 2 main consumer apps, Warpcast and Supercast, let user report casts, and hence are able to identify signers that have a potential bad behavior. The proposal above works for them as well.
Does that really matter ? 🤷♂
Yes, there are bots in DAUs. But those who are raising that as a massive problem, and pointing that as a source of "fake" growth have yet to do the math and prove their point.
The fact is that around 97% of messages come from Warpcast or Supercast signers. Both being end-user consumer clients, chances are these messages are not from programatic bots.
Measured vs perceived quality 💎
DAUs are quantitative figures, that look globally at the entire network.
Nobody can, with human eyes, perceive the entire network figures. With less than 1000 followers, and a very low ranking in the $DEGEN game (ie low allowance), I am not the target of most bots, and I see a rather qualitative growth.
Big players, with 100k followers or more, some big money or blue chips on their verified addresses, will obviously be a better target for bots looking for attention and validation.
As said above, looking at the data analysis from hubs, if there are "programatic" bots, they are a minimal part of the DAU. But they might stand out more in front of users with a big following, making them more visible.
Growth trend vs spot value 📊
When using KPIs, focussing on the spot value is most of the time narrowing the vision and the lesson they teach us.
There is no perfect metric, or if there is one, and calculating it costs more than building a good product, you might not want to focus on it. Having a good enough, easy to collect metric is a great way to consistently measure growth or trends.
If bots were invading Farcaster, users would be running away faster than bots are coming, and ultimately the DAUs would go down hard.
Keep building 👷🏻♂️
Despite potential discrepancies, DAUs remain a useful metric for identifying trends if used thoughtfully. As the Farcaster community grows, it's crucial to maintain a balanced view, focusing not only on quantitative figures but also on the qualitative aspects of the platform's development.
There are many options to make it harder for spammers, scammers, farmers to bot the network, and they all come at a cost. But we should only try to solve the problem if there is an actual problem.
If we keep building, a day will come when indeed the Twitter / X professional bot-makers will try to settle on Farcaster. This will be a great day and the entire community will be ready to fight with their best weapons.
If you want to be one of the firsts beta testers of F4R, the social media marketing platform for Farcaster, subscribe !
