To re-iterate on what we're trying to do here:

We're trying to highlight the exciting parts of Ethereum that make people want to wake up in the morning and build for this ecosystem.

Just to be super clear here:

- A cheap-effort merch store where I can buy a "Built on Ethereum" cap with Google Pay is not it. I have removed that link twice already. I think you as curators do have the capability of not submitting that. Let's please not submit every crappy thing. We can all be opinionated and submit high quality stuff.

- A website where I can do an onchain DeFi thing marginally better than before, is also not that. Screw Base and Unichain. They are inherently NOT interesting. I'm bored like crazy just based thinking of what they want to do. I'm not saying that such submissions aren't useful or "full of utility". But seriously people: Nobody fucking cares. Why would I care about doing an onchain thing 5% better? I'm literally going to quit Kiwi News if this is all there is to Ethereum's potential, it's soooooo boring!!!

You know what we want: We want to hear about the Ethereum visions beyond paper-clip-optimizing some DeFi transactions. We want to hear about builders taking risks and we want to hear their stories.

We want to create a cultural revolution in the space. Ethereum has been attacked from all sides. It has been canceled, some have been put into jail, etc.

Yes, we're saying that we don't like that KOLs are putting paper-clip optimizing ETH researchers on a pedestal. At the same time, it is on us to actually find the positive examples of building the Ethereum visions. We need to find those as they give suppression fire to those developers who want to build exciting visions but who fear for being attacked.

Btw, I often get the impression here that I say something in this community and I try to say it as the leader.

And in my view I communicate this sufficiently clearly. E.g. even with LLMs I now have a counter factual where I can say something and in many cases the LLM actually understands me really well. So this would lead me to believe that it isn't my communication capability that makes me fail to generate alignment. But then somehow I often feel like that comfort or dissent actually trumps my mandate here in the community.

In my view we have to find greater alignment otherwise we're not going to succeed and this site is going to fail like any other.

I know the crypto space is all sorts of bullshit about decentralization, no censorship, no leader, power and whatever. But I feel like we're going to have to really step up our game in terms of aligning wrt goals etc. because otherwise we're not going to go anywhere.

I see our trajectory as a vector. If I wanna move in a certain direction, and others want to move in different directions, in the crypto space I only have soft power influence. We're peers here, I can't pressure anyone really.

So it is extra important that we step into conflict and we find out where our commonalities are. Otherwise we're going nowhere.

I often get the impression that comfort or passive aggression is the norm instead of alignment: "I can just do my own thing without communicating that I'm unaligned right now." or "No, you the leader, you are wrong and I'm just going to insist on doing my thing"

I think this is really bad and should stop. Otherwise we can't scale and we won't be able to monetize this thing here.

E.g. even with LLMs I now have a counter factual where I can say something and in many cases the LLM actually understands me really well. So this would lead me to believe that it isn't my communication capability that makes me fail to generate alignment. But then somehow I often feel like that comfort or dissent actually trumps my mandate here in the community.

Idk, when I speak with LLMs about a more open-ended subjects (like content curation) and give him instructions, it still often gives me something I don't want. Or makes mistakes.

Sometimes it's because my prompt wasn't clear enough, sometimes it's because it needs more data for edge cases, and sometimes it just isn't capable of giving me the type of answer I want.

Just like humans.

My thoughts about content are that:

1) We had a certain content policy for almost 2 years,

2) We announced a big change in this content policy,

3) It took some time for people (including me) to get a better feeling of what we're looking for,

4) We are seeing fewer unaligned content pieces: you can even see it in our moderation file when you compare how many links got moderated before and how many are moderated today,

5) This would suggest that the trajectory is right, and maybe we just need more time and feedback,

6) Maybe we would never reach the perfect state where every submitted link first perfectly into our content policy

7) But 6) doesn't mean that we shouldn't aim for that

OK sure, we need more time.

That said, I think actually an LLM would be capable of potentially even better curating the content on the site if we managed to come up with a really narrowly designed document to filter what we want.

What we could do is to collaborate on refining the prompt. The LLM would then read the content, and it'd come back with an answer on whether this complies with what we're aiming for.

For humans this process would also be useful because it'd as well be an alignment exercise.

Idk, when I speak with LLMs about a more open-ended subjects (like content curation) and give him instructions, it still often gives me something I don't want. Or makes mistakes.

skill issue? That aside, I actually do think that LLMs are extremely aligned for process outcomes where "once you see it, you know if it's right or not"

And content filtering at least is such a thing.

That said, I have also been positively surprised by how some curators are submitting seemingly unrelated topics and then manage to connect the dots to Ethereum. I'd be happy to see more risk taking on that end.

Makes me think of the Botto approach, where the AI artist, or in this case LLM would supply content outputs and human curation would guide it (Botto is also learning from the curation). The thing is to get it to work well you need quite some human engagement, it is also something Botto has to push for all the time. Bottom line is always how to get people’s attention, how to get people to play? If you want attention, you’ll need some level of playfulness or game. Ideally there has to be something at stake. But I think it’s good for Kiwi to think out of the current Kiwi box. It is going to be hard for an open system how it is set up now - anyone can buy a Kiwi nft and become curator - to have full alignment. In fact in a sense anyone can just post anything. I think in that sense it is going quite well. You could have had a ton of spam instead.

That said, I think actually an LLM would be capable of potentially even better curating the content on the site if we managed to come up with a really narrowly designed document to filter what we want.

Yeah, but then we would need to write down the guidelines which I have been advocating for some time :) A precisely written document like that is useful for both LLMs and humans, especially if we provided examples of content we like and don't like.

I generally agree that AI here can help us a lot, it's been helping YouTube, Facebook and other social networks for years. I imagine it could also read comments to check if the dot connection to Ethereum has been made and if it makes sense.

If you want attention, you’ll need some level of playfulness or game. Ideally there has to be something at stake. But I think it’s good for Kiwi to think out of the current Kiwi box.

100%

I think we made a mistake by not continuing to develop Kiwi Karma as a currency. I also think there is currently nothing that really aligns the Kiwi community. If we had a token we'd all try to make the token number go up or whatever. At the moment, IMO, our interests are too diverse, dare I say, pluralistic, to go in one direction together.

I think in that sense it is going quite well. You could have had a ton of spam instead.

It is true. But I think if we had a lot of spam, at least that would signify that it was really important to be seen on Kiwi. If you think about it, at the moment, Kiwi is so unimportant that we don't even have spam.

I don’t know exactly what 5% DeFi thing you are referencing but I generally like to see new apps or even a 5% twist on apps, primarily because we are all building on top of each others innovations, so even if it’s not seemingly interesting it can spur an idea for a community lending program, etc. But in a lot of ways it depends on the product. Maybe the key is not a link to the product, it’s not product hunt, but someone talking/analyzing the product, idk. LLMs do you understand what I mean?

Re:LLMs

I’m testing a new bot from Coordinape in the /purple channel. It reads every cast and if the cast contains Purples values, it automatically gives you GIVE (not a token an eas attestation). The goal is to give reputation to community members who post insightful on topic things. The bot is programmed with purples values which are around decentralized social, farcaster protocol growth and improvement. Importantly I wanted to make sure it was just all bs casting so posts that are critical of Farcaster should also be rewarded (have not seen in prod yet).

/purple is low volume so I’m looking to expand to other channels if anyone has a suggestion.

Maybe the key is not a link to the product, it’s not product hunt, but someone talking/analyzing the product, idk. LLMs do you understand what I mean?

Yes, maybe you're right, this would make a difference