avivo

I think if it ultimately protects customer data in a significant way, I would be for it.

Are you able to share any more detail on how you determined this is the best route? It would be a significant implication for many other pieces of open source software also if so.

(And I say this is someone who just recommended cal.com to someone a few days ago specifically citing the fact that it was open source, that led to increased trust in it.)

I did find the video valuable, for reference for others: https://www.youtube.com/watch?v=JYEPLpgCRck

I think if you are committed to switching back to open source as soon as the threat landscape changes, and you have some metric for what that looks like, that would be valuable to share now.

I would like to see the analysis that you're referencing around open source being 5-10x less secure.

Open source is one of the main reasons I recommended cal.com to everyone — I just did so yesterday again in fact!

I'm disappointed to hear this especially since I don't think the rationale makes sense, from what I understand of the security landscape, and it also makes me a little more skeptical of cal.com in general.

They could also buy potentially Zulip, an OSS slack alternative with a much better conversational model.

There is a lot of potential in this. Not just for government democracy, but for also introducing democratic elements into tech/AI policy, when that tech has impact comparable to many governments.

I worked in tech, and after some formative experiences, shifted to working on helping ensuring ensure that tech's impact on society can serve the public interest. But that leaves the question of what "what is the public interest"?

Sortitition / lottocracy / deliberative democracy / mini-publics all roughly refer to the same way to answer that question — providing a representative microcosm with the space to deeply examine an issue, and come to a set of recommendations and decisions on it. Unlike with electoral democracy, it's faster to spin up and experiment with, and it's harder for bad actors to entrench power (elections can be useful, but they're one of many tools in the democratic toolbelt).

That thinking lead to https://www.belfercenter.org/publication/towards-platform-de... . That basic approach, has been somewhat picked up by Meta (https://www.wired.com/story/meta-ran-a-giant-experiment-in-g...) and Open AI (https://aligned.substack.com/p/a-proposal-for-importing-soci..., ~ leading to their democratic inputs and collective alignment work).

I've now started an organization focusing on applying this and other democratic paradigms to decision-making about AI (https://aidemocracyfoundation.org/) as a way to solve a variety of challenging governance problems across the AI stack. If you're curious about it, our ICML paper goes into more detail: https://arxiv.org/abs/2411.09222 .

The FTC also recently put out a statement that is fairly pro-open source: https://www.ftc.gov/policy/advocacy-research/tech-at-ftc/202...

I think it's interesting to think about this question of open source, benefits, risk, and even competition, without all of the baggage that Meta brings.

I agree with the FTC, that the benefits of open-weight models are significant for competition. The challenge is in distinguishing between good competition and bad competition.

Some kind of competition can harm consumers and critical public goods, including democracy itself. For example, competing for people's scarce attention or for their food buying, with increasingly optimized and addictive innovations. Or competition to build the most powerful biological weapons.

Other kinds of competition can massively accelerate valuable innovation.

The FTC must navigate a tricky balance here — leaning into competition that serves consumers and the broader public, while being careful about what kind of competition it is accelerating that could cause significant risk and harm.

It's also obviously not just "big tech" that cares about the risks behind open-weight foundation models. Many people have written about these risks even before it became a subject of major tech investment. (In other words, A16Z's framing is often rather misleading.) There are many non-big tech actors who are very concerned about current and potential negative impacts of open-weight foundation models.

One approach which can provide the best of both worlds, is for cases where there are significant potential risks, to ensure that there is at least some period of time where weights are not provided openly, in order to learn a bit about the potential implications of new models.

Longer-term, there may be a line where models are too risky to share openly, and it may be unclear what that line is. In that case, it's important that we have governance systems for such decisions that are not just profit-driven, and which can help us continue to get the best of all worlds. (Plug: my organization, the AI & Democracy Foundation; https://ai-dem.org/; is working to develop such systems and hiring.)

Author here, not of the post, but the linked article, making the case for bridging-based ranking, https://www.belfercenter.org/publication/bridging-based-rank... and the related https://bridging.systems/ which lays out a set of open problems and approaches.

It's great to see this exploration. Those interested might also want to check out https://vitalik.eth.limo/general/2023/08/16/communitynotes.h... and https://knightcolumbia.org/content/the-algorithmic-managemen... (and how aspects of this might be applied to AI governance https://reimagine.aviv.me/p/governance-of-ai-with-ai-through ).

If you can provide a script that takes in an HTML file and provides an image ready for rendering, that would be amazing. Then I can automatically take any website and have a cron job that dumps the result into a shared dropbox link where it can be used by the screen.

It's worth knowing that, according to some recent research, taking a topical antibiotic like this might have permanent impacts on microbiome in your nose (and lead to drug resistance.)

What is the advantage of this versus running something like https://github.com/simonw/llm , which also gives you options to e.g. use https://github.com/simonw/llm-mlc for accelerated inference?

Fascinating, this led to this projects which is pretty interesting. https://www.modos.tech/blog/modos-paper-monitor Seems like a very exciting effort and product but I'm not sure if it's still active.

Has anyone gotten this to speak back a response on a Mac? I don't see a setting for this.

I'm also curious if there are ways to do this for Google Home and Alexa.

How are people creating transcripts like this? There doesn't seem to be a convenient export or save option for conversations.

It's worth understanding the community and org better, and their reaction. Relevant links:

- https://meta.stackoverflow.com/questions/421778/how-do-you-p...

- https://meta.stackoverflow.com/questions/412696/is-it-accept...

- https://meta.stackexchange.com/questions/384355/could-chatgp...

I think this gets almost all the way there but not quite — there is one more vital point:

How we act depends on our environment and incentives.

It is possible to build environments and incentives that make us better versions of ourselves. Just like GPT-3, we can all be primed (and we all are primed all the time, by every system we use).

The way we got from small tribes to huge civilizations is by figuring out how to create those systems and environments.

Yes, the algorithm is not the problem alone, but a good algorithm can help fix the problems — since it creates the "loss function" (the incentives) for the humans using the platform (I go into that in more detail here https://twitter.com/metaviv/status/1529879799862378497 and here https://www.belfercenter.org/publication/bridging-based-rank... for those who are curious).

So it's not about "reaching for the stars" or complaining about how humanity is too flawed. It's about carefully building the systems that take us to those stars!

This is on point

The goal of all this activity is not to debate, converse or exchange information. The goal is to win by being maximally controversial, as that's the behavior that is rewarded.

the real issue: what gets amplified

But it can be (at least partially) fixed if you change the optimization function.

I advocate for that here: https://www.belfercenter.org/publication/bridging-based-rank...

And Twitter's Birdwatch (that Elon recently got all excited about when it fact checked the White House: https://twitter.com/metaviv/status/1587884806020415491) actually does this "bridging-based ranking" for adding context on tweets.

Here's the paper with details on how it works for Birdwatch: https://github.com/twitter/birdwatch/blob/main/birdwatch_pap... (you can also check out the source code in that repo).

Halfbaked thought: Have it cost half the amount as normal mail to receive email there. Perhaps also gov use is free, and users can whitelist specific domains or subdomains to be free.

It turns out that we like people the best when they respond to us the fastest––so fast (mere milliseconds!) that they must be formulating their reply long before we finish our turn.

This might be true; but looking into the linked study, it appears to be on Dartmouth students. This claim at least maybe culturally dependent.

Can you install a real keyboard tray?

I'm curious if Astro supports git-backed visual editing? I can't seem to find that, but it would rather useful if so! (ala tinacms.io , stackbit, etc.)

It looks like it provides out of the box support for Overleaf! (Collaborative LaTeX editor) https://www.overleaf.com/blog/635-languagetool-a-free-browse...

This is not true of Grammarly last I checked, you have to hack something together... (e.g. https://medium.com/@tardijkhof/how-i-made-grammarly-seamless... )

I think what would be most interesting and helpful is an augmented form of reading that is semantic. I would love to have bold and italics and even section headings and table of contents that are toggleable on and off as I read anything, to quickly skim and focus on the most relevant content.

I can also imagine a version of this which is contextual, based off a query, or a personalized recommendation system.

I spend much more time skimming than reading, and skimming to determine if something is worth reading. Anything that can support that is incredibly valuable and would increase my functional reading speed for accomplishing tasks.

Fascinating. "Deprank is particularly useful when converting an existing JavaScript codebase to TypeScript. Performing the conversion in strict PageRank order can dramatically increase type-precision, reduce the need for any and minimizes the amount of rework that is usually inherent in converting large codebases." I wonder if the idea of using pagerank style systems for ~refactoring translate to other domain; e.g. organizational or knowledge refactoring (ala https://aviv.medium.com/when-we-change-the-efficiency-of-kno... )

Not exactly right anymore. Check out https://www.unepfi.org/publications/investment-publications/...

"Large institutional investors are, in effect, “Universal Owners”, as they often have highly-diversified and long-term portfolios that are representative of global capital markets. Their portfolios are inevitably exposed to growing and widespread costs from environmental damage caused by companies. They can positively influence the way business is conducted in order to reduce externalities and minimise their overall exposure to these costs. Long-term economic wellbeing and the interests of beneficiaries are at stake. Institutional investors can, and should, act collectively to reduce financial risk from environmental impacts."

Also https://theshareholdercommons.com/ / https://engine1.com/ that are working to embed this into corporate DNA

It is a choice at the software later. And who is making that choice? It's the initial people who actually write that software.

But what if that system is now affecting many other people, or the entire planet in a significant way? Should they have some voice over that?

Under a traditional governance regime the answer is that that is at least possible to change. It may be difficult, but it does not violate the laws of physics and can happen in less time than the heat death of the universe. But we can now write software that makes it functionally impossible for anyone to make that choice, even potentially the original designers. That is an option that we did not have before. It's in some sense the essence of trustlessness.

In some cases, this might be the right trade-off. For example, beyond the blockchain, this is also a way to think about encrypted communications. It is a very significant new power that we can now wield.

But it must be wielded carefully, and that doesn't seem to be happening.

So yes, blockchain applications don't need to be irrevocable. But the ability to make them so is something that could have a very significant implications—potentially negative.

As a somewhat tongue in cheek example, but with a little bit too much reality to be comfortable, this irrevocability might allow you to "create" a paperclip maximizer DAO (incentivized at the social layer, with humans doing the work).

It's a good question. There are several kinds of answers, two of which I allude to but don't go into a lot of detail. The answers based off of peer reviewed research can be found in the article in Science that I quote. There is also the answer based off a sort of existence proof—so far, even when this is been done at a high level around extremely controversial and impactful issues, this has not appeared to be a significant problem.

To be clear, the process that I am describing there is not some newfangled idea that I just dreamed up. It's something that has been used by France, Ireland, South Korea, even the EU as whole, is likely to be institutionalized as part of the governance of these organizations. It's just fairly new so not that many people have heard about it.

There is also a rebuttal article by the same fund: https://www.bvp.com/atlas/the-antidote-to-cryptophobia/

That said, I think Adam (the original article) is closer to being right.

He doesn't even get into the thing that originally made me incredibly excited about the potential of blockchain tech—the way it lets one create a new sort of custom and irrevocable 'physics' for information and incentives. But the same irrevocability is now what makes me deeply concerned. If we get something wrong, it may be impossible to change, unlike normal human systems. It's like building Facebook, except the original design might be at least partially locked in functionally forever—and so issues created by naive founders can never be resolved and may warp our political and economic systems.

I have also generally found that crypto/blockchain/web3 doesn't address the problems its adherents say it can solve ( https://aviv.medium.com/the-magical-decentralization-fallacy... ), particularly changing power structures to get to a better world. Trying to find alternative solutions to address the power issues around centralized platforms is what eventually led me down a very different path ( to http://platformdemocracy.com/ ).

This is one of those brilliant little things that I just want to see everywhere.

This is great. That said, I would love to see this extended beyond just keyword search, to contextualization of a given piece of content. https://aviv.medium.com/contextualization-engines-can-fight-... ( I should note that the word authoritative here just means a particular 'lens' that has been vetted by a defined process—it's a way to make the problem simpler for prototyping. It might for example be a lens showing just accepted Stack Overflow answers.)

Imagine being able to select a bit of code in your editor, and getting an explanation of what that code does based off the content of all of the accepted answers on Stack Overflow.

There actually is a way to give devolve power away from both the government and private companies. Details here: https://www.belfercenter.org/publication/towards-platform-de...

Essentially get a representative sample of the population, pay them and train them up (through neutral 3rd parties), and have the make the ultimate calls.

This isn't a crazy new idea; it's been done around the world for many complex and controversial issues. Here's a video from the Economist on this https://www.youtube.com/watch?v=6UcFQ-eDhTk and here's examples of uses of this approach around the world https://www.oecd.org/gov/innovative-citizen-participation-an...

Ease of access does matter.

It only buys time, but that can provide the time needed to create countermeasures and ideally make those very accessible—somewhat similar to responsible vulnerability disclosure.

This piece goes into more detail: https://aviv.medium.com/the-path-to-deepfake-harm-da4effb541... (excerpt from a working paper, part of which was presented at NeurIPs).