Offers unchained: Federated Offerbots

I want to draw your attention to a discussion taking place on our Humane Tech Community now. One of our members @AndrewMackie has published a brilliant piece of work on the mechanism of Offers, and why they cause such big problems with aggregators (FAANG et al).

His problem analysis is a very refreshing, clear-headed approach that avoids the use of big, alarmist and emotionally loaded language you usually hear in the field.

But its the solution side where Fediverse / AP could perfectly fit in. In daily life offers are the most natural thing. With the ambition of the fediverse to replace the broken web, eventually it will have to provide good representation of how that works online. In other words, before commercial interests start to take note of the fediverse, throw their money at it to create another mostrosity, the fediverse should have figured out a decent enough solution to handle Offers with most loopholes covered.

There is some urgency in this, I think.

I recently started to elaborate two fediverse projects as part of my innercircles (teaser only here) experiment, and with Andrew’s involvement, we’ll investigate Offers further. But we’d love to see that become a broad-based effort.

Here’s the discussion: Offerbots (to bypass aggregators of attention). Would be lovely if you join there, and give Andrew feedback on his work, but here will do fine as well, of course.

PS. innercircles will be open, non-profit and foss and hosted on Codeberg, where currently the delightful project is already part of it. Still in preparation, but if you want to know more you can join our humanetech forum and DM me there (@aschrijver, community facilitator).

(Photo by Savvas Stavrinos from Pexels)

Hey, thanks for sharing! Really interesting!

I like the explanation of the Offerbot model and the pictures. I can tell there’s been a lot of serious thought and attention paid to framing the problem happening on a social scale.

I am concerned about the lack of tie-in to the wealth of existing research into the theory of the attention economy, even the footnotes do not tie into that existing research. It does reference a paper and a quote from Marissa Mayer, which is a start. However, the lack of tied-in research to past efforts/research (and showing how the past efforts weren’t able to solve this, or didn’t account for new research) makes me afraid that this is one of those concepts that is developed in isolation from other ideas, and hasn’t really been weathered. That is worrying to me, because there may be problems with the problem description. And I also cannot judge how receptive work is to alternative ideas.

I also think it undermines its own credibility at times with political statements like: “It should not surprise us, then, to discover that aggregators have harmed our economy, society and information.” Such statements are not useful to the previous descriptions and only serve to alienate or virtue signal, as people can still support the Techno-Kleptocracy model w/o the political statement. Especially considering an Offerbot is, in a fit of irony, just an aggregator at smaller scale – surely the author does not intend to also mean Offerbots are harmful!

I also think at times the problem statements go into outlining too many effects, without breaking problems down. This kind of approach is great for motivation and getting like-minded folks enthusiastic, but I don’t think it’s a good-enough launchpad for developing engineered solutions. For example, there’s a whole page on how capitalism and free markets work, and then some application to aggregators. From there, it wants to jump directly into solutions. I think there’s at least 2 major pieces missing:

  • Which specific societal effects can be addressed by what kind of specific technological properties
  • What past solutions have done in this space, and why they are insufficient

The reason why I am worried about the first bullet point is that, to me, it risks falling into the “well-intentioned but misguided” bucket. Since the problem is overly-generalized, and yet specific technological concepts have already been mentioned like key-value pairs, I’m afraid it will run into a lot of muddy and unproductive territory. “Why does it need to be a KV pair?” “Why JSON-LD signatures?” etc. It also does not acknowledge that some problems are just human in nature, not technological. For examples, see: BitTorrent, spam. It would help if there was an additional page breaking down the specific existing problematic properties in the attention economy, identify desirable alternative properties, and only then that specific solutions tie in to this. For example, alarmingly, “Privacy” is a small “Other Problem”! No one is going to disagree with a generic shout of “for the sake of Privacy”, the devil is in the details.

The reason why I am worried about the second bullet is I am afraid the solutions are going to be a “NIH” syndrome or “reinventing the wheel” or other solutions will be rejected if they don’t fit into the Offerbot philosophy (see previous fear about openness to alternatives). Furthermore, there’s the real fear that solutions are chosen prematurely or without understanding the full extent of the problem (for example, shouting “Federation” without even mentioning Embrance/Extend/Extinguish [EEE] that led to the current situation). Taking this consideration with the previous one has me really worried.

I bring this all up because I am interested in evolving the ActivityPub spec to address some very specific concerns in the general categories of Privacy, Security, de-platforming, and increasing resistance to EEE. The past few years I’ve had some conversations w/ folks around this and there seems to be others that have really great understandings of very specific problems around these broad topics & how they relate to AP, and I’d like to make it a reality in the near future. It would be a shame if Offerbots and the-next-AP-version had different concrete technical approaches (for example, JSON-LD Signatures is not a popular topic of discussion) which led to fracturing of the ecosystem.

Disclaimer: I am an employee of the G in FAANG but everything I do in the ActivityPub world here is entirely me and my own work and not a reflection of my own employer and I do not speak for them, I figure this transparent reminder would be appreciated on this topic since I can’t remember who all in the community know knows/cares from when I mentioned it at the FOSDEM roundtable.

1 Like

This is some wonderful and extensive feedback @cjs and I thank you a lot for the time you spent writing this down!

I think Andrew will be quite happy with this response as it consists many handholds for improvement. Though I can’t speak for him, of course, I think you see most of your arguments correctly. But that is exactly the reason why Andrew has decided to put his information out to the world, realizing he could not take this further alone.

And that same exact reason is why I posted in this forum. To collect expert advice like yours. To find pointers to existing bodies of work. To motivate others in their own projects to come forward, cooperate, etcetera. For my own use case - which is very small in scale and not directly fediverse-related on the Offer part - I’d like to see what can and cannot be incorporated in a more or less standards-compliant manner.

I am not sure how best to follow up on all the points in your feedback, but a suggestion is:

  • For me to either copy/paste to HTC or you to join and re-post (I’d gladly welcome you), so we can go into more of the societal aspects of both problem and solution space.
  • Then to have the more technical discussion here, for all the parts that are most relevant to the fediverse.

Thanks again!

PS. A bit of background on Humane Tech Community and my role there:

I formed the community as volunteer, after Tristan Harris et al did not have time to steer the ‘Cultural awakening’ they promised when founding the Center for Humane Technology. Their reason was a good one: their extreme success at reaching highest levels of the tech world and political realms (world leader level).

Having been very active as initiator and facilitator of HTC - now affiliated but completely independent of Tristan Harris’ CHT - I have become much less active for the last year or so, and so too has the community as a whole. Reason was I became bone tired raising awareness of problems, seeing others do that too in various ways, always regurgitating the same talking points, while showing precious few focus on solution-finding. Last act before my relative inactivity was a forum reorganization along the ‘solution-axis’, creating a community website and me defining a framework called The Pyramids of Humane Technology in my LibreOffice. That is still what it is, it has no hands, feet, nor teeth as yet. But I have indirect plans to give it more body.

I promote both fediverse and humanetech wherever and however I can.

Thanks @aschrijver and hello @cjs, and thank you for your feedback.

The offerbot concept has been developed in isolation, not by choice but because I struggled to communicate it clearly to people in the past and, in particular, to explain why offer processing is the key to everything else. In the past I have contacted many people asking for input and help, from TBL to linked data experts and back again. So I did it on my own. Now that I have a website which explains my understanding of the problem, I am sharing that with people and will hopefully make more progress, aided by people like @aschrijver, thank you.

My concern with other research on the attention economy (e.g. Tim Wu) was that it focused primarily on privacy and legislation-based remedies, not the distortion of markets and the replacement of the mechanism for making and receiving offers, which was my primary focus. What research would you suggest that I consider?

I’m a little confused by this.

Firstly my sentence about harm is merely an intro and hook into the next article (where I make further arguments).

Secondly, I’m confused about your (apparent) acceptance of the Techno-Kleptocracy model but rejection of the harm of acting like a Techno-Kleptocrat? Consider that Google, for instance, ranks every product and service by popularity (a single heuristic) and is therefore a kingmaker, blessing popular vendors with more trade. It also creates an underclass of unpopular vendors who must compete in an AdWord’s auction for Google’s blessing, allowing Google to act as a rentier. Google is, over the long term, making the rich richer and the poor poorer (and Google the richest of all). Is there no harm in doing this?

Offerbots are not an ‘aggregator’ as defined by the website, where I am using the term to refer to companies who aggregate. Certainly offerbots aggregate offers, but where FAANG aggregators do that to capture attention and sell it to the highest bidder (by selfishly selecting the set of offers displayed to users to maximise the likelihood of payment to the aggregator and maximising the amount paid by the vendor), offerbots give their owner control over which offers they see and consider. It’s an information economy, not an attention economy.

The harms to privacy are well documented elsewhere. What I am bringing to people’s attention is the harms to our economy, society and information (which are ignored elsewhere).

On the technology, in recent years I have become a full-stack developer but I am not a technologist. For me technology is a means to an end. I am happy to consider any technology, noting that the distributed systems have thus far failed to capture the imagination of the public and, therefore, we should not presume that simply building it in adherence with existing standards will make the public come.

[Edit: It’s worth noting that offerbots must be highly efficient at sending, receiving, processing and representing offers in a generalised manner at a reasonable scale. It may not be possible, much less desirable, to shoehorn offerbots into technologies which are designed primarily around storing personal data if doing so compromises the core function of the offerbot.]

Thanks again.

Thanks @AndrewMackie for taking the time to respond here! I understand the struggle to communicate clearly all too well – as you yourself pointed out, I have some clarifying to do!

Starting off, I want to reinforce that I really do appreciate the time and effort you’ve done to create the images and put into words your ideas. I hope I can clarify that my criticisms are not meant to be rejections and recognize I tend to have a bit of an academic-y bias. So I hope my feedback helps you feel like your site’s even more compelling.

Onto your response…

With regards to Tim Wu, your point is interesting. If you choose to expand the privacy section, I don’t think it would hurt to relate your ideas and his. Noting commonalities and differences, even if the Privacy section is short, would strengthen it greatly: you’ve identified a weakness in his work (you’re more market focused) and you’re filling in his gap. Or once you’re the next generation famous activist… he’s filling in yours. :slight_smile: Additionally, some problems are more human than technological in nature, and that is where the legislation remedies may need to come into play. Your existing work definitely focuses on market forces, and for separate reasons I would think it would benefit greatly breaking some of those properties into specific problems. I suspect you may find some problems have technological solutions while others not as much, in which case you have another opportunity to relate back to his work. Big grain of salt: I say all this but I haven’t read particular books of his, I am just familiar with some of his activism.

One particular work I would love for you to ingest and reflect on is Michael H. Goldhaber’s “The attention economy and the Net”. This is an older piece from 1997 but includes ideas around wealth and property in the attention economy, which I don’t recall reading about in Offerbots. The specific ideas may not have aged well, which is a great opportunity for you.

As for the political statement, I think I definitely caused confusion here.

In my mind, there are two ideas I read in your work:

  1. There is a Techno-Kleptocracy (a centralizing of technology) and they own aggregators
  2. Aggregators are harmful for the economy, society, and [quality of?] information

When I said “support the Techno-Kleptocracy model”, that was poorly worded. I intended to mean that the two points above are separate beliefs. It is possible for a person to believe in both points, it is possible for a person to believe “there is a centralizing of technology, but aggregators are not inherently harmful”, it is possible for a person to believe “anything that aggregates is harmful, also technology is still decentralized”, or neither.

I understand we may have a personal disagreement here: I don’t believe there’s anything inherently harmful with aggregators. A concrete example: I don’t think the aggregating technology behind DuckDuckGo’s Search is harmful. I do hate to see a centralization of technology+data though – to me that is the potential root-cause for negative effects, regardless of the software category (your aggregators idea or otherwise) – so I’m in the “point 1 agree, point 2 disagree camp”.

Having said all this, my specific feedback is that in an article all about point 1, it ends with a sentence on point 2. Which was a bit of a carpet-pulled-out-from-under-feet moment for me, felt alienating, and raised some questions. Depending how you want to handle this (not handling it is also a fair choice), I’m happy to elaborate more, if needed.

This does lead to the discussion over aggregators and its definition. I think we have differing ideas here – not the nitpicky semantic kind – as I am a firm believer in “people have finite time which means the computer has to do some sort of incidental censorship (via filtering, aggregation, prioritization, etc)”, and acknowledge that who has that control matters, and acknowledge that users will at times be willing to hand that control over as exercising that control also takes time and energy. As I type this out I’m going on a tangent – I think I’ll drop this feedback. I understand how and why you do not view Offerbots as an aggregator, and I’m not here to change your mind on that.

On the “small privacy section” bit, I agree it may be documented elsewhere. I hope prior paragraphs cover my reasoning for citations/expanding it.

On your final point: Sounds reasonable. I hope the Fediverse is a good fit, but I’m not optimistic based on what I understand so far. Unfortunately it is currently susceptible to the same EEE that has led to centralization of previous technologies that were once-upon-a-time more federated.

Thanks for taking the time to read!

1 Like

I truly love :heart: this wonderful exchange. I don’t have anything to add on this topic atm, but I created a topic for the EEE tangent:

[This is a repost which, I hope, will be easier for you to reply to, so that we can, if you wish, have an ongoing conversation. It also contains a couple of ideas which were not part of my last reply. As a new user I’m not allowed to reply to my own reply, so I had to wait 24 hours for the old reply to be deleted …]

Thanks @cjs, I appreciate your feedback and am incorporating it into my website - it will not be wasted.

For example, I have changed the sentence on the harms of aggregators which you found to be jarring (to make it a question rather than a statement).

I’m also trying to concentrate on the why and how (which, having studied the problem at length I am best qualified to answer) rather than the what (the specifics of the technology, which I am not). To that end I have started writing another article: The Purpose of Offerbots

To your most recent comments:

To clarify, are you suggesting that I talk more about privacy on the problem side or the solution side (or both)?

Also, something occurred to me about this which might be instructive.

I believe that the Fediverse, Solid project, etc.(and possibly yourself) consider privacy and data ownership to be THE problem which must be addressed by technology … ?

I, however, I see billions of people voluntarily giving away their privacy and data to obtain offer processing from FAANG and other aggregators. I believe that the real problem to be solved is providing individuals with sovereignty over their own offer processing (in a way which also delivers privacy and data ownership). Privacy and data ownership is absolutely critical, but it’s a second order problem (and the only way to solve second-order problems is to solve the first-order problem). I have updated ‘The Problem’ page to reflect this (and welcome your feedback on it).

This also explains, to some extent, why the offerbots project was conceived in isolation. Each project exists in its own world - it has its own view of the problem and the solution. To understand whether Solid (say) is a good platform for Offerbots requires immersing myself in the Solid world for a long time, to understand its origins, its trajectory, its technology, its maximum scope and so on.

If I were to commit to building Offerbots on Solid, it puts my project at risk of EEE from the Solid community if I’m unable to convince them to change the priority of their project (so that offer processing is the first-order problem and privacy and data ownership are considered second order problems).

It’s also, no doubt, why TBL didn’t respond to my feeble attempt to contact him (because jumping into my world from the Solid world is difficult, particularly if he didn’t agree, from my feeble attempt to explain the problem, that the problem he was addressing was actually a second-order problem).

And yet I am constantly learning what I can. As an example, when I started the project (before I gained skills as a full stack developer and, therefore, before I understood what the web really was and how it worked), I thought that offerbots were something different to the web, that it was going to require a new communications protocol other than HTTP. But then some Linked Data people (Tim Holborn in particular) explained to me that the web can do what I want (and now that I build web apps, of course, it’s clear to me as well).

So it may be that Solid (or something else) is the right platform for Offerbots, but I don’t have the bandwidth to consider all of them. What I need is help from people who are immersed in those technologies to consider the claims in the offerbot project problem statement (the offer processing is the primary problem), help me refine the solution and then work out which existing technologies can be built upon (rather than inventing the wheel).

I have read that article in the past. It makes some significant errors regarding the value of attention, the most significant of which is suggesting that attention is the valuable resource in and of itself rather than being valuable as the medium for making and receiving offers in order to trade.

If the article was correct, a farmer would be willing to give me a sack of potatoes in return for me giving him my attention (perhaps to tell me a story that no one else wants to hear). The farmer certainly wants the attention of his friends to tell those stories, but the farmer wants my attention solely to communicate that he’s selling bags of potatoes at his front gate for $5. That’s because he wants my $5 in order to save up for a new tractor (which my attention will never buy him) and he doesn’t want to waste his attention gaining my attention for anything else (as he has to give his attention to growing potatoes!).

Aggregators cannot, therefore, just capture the attention of random human beings and sell it to other random human beings. They must capture the attention of potential buyers for a particular good and sell it to vendors of that particular good because those vendors wants buyers’ money and not their attention. The attention is the medium for the exchange, not the valuable good being exchanged.

The article suggests that any economy which exists in cyberspace cannot be based on material goods (because cyberspace is not material). By that logic, television advertising never sold a physical product to anyone. Were television advertisers of the 1950s attempting to sell virtual goods to television viewers? It’s not a sensible argument.

It also says that money will become less important as currency, being replaced by attention. But attention cannot be stored, it is not fungible and therefore it could not be a unit of value (even if we valued it in the way that Goldhaber suggests).

As I said on the other forum, the purpose of offerbots is to shift us from an attention economy (by which I mean being given ‘free’ information processing in order to capture and direct our attention toward advertising) to an information economy (in which we pay for our own information processing in order to control which offers our attention is allocated to).

Thanks for clarifying. I don’t know what articles you’ve read of mine with regard to The Problem. If you didn’t start at Offers and Attention and work your way through External Offer Processing and Aggregators then you won’t have the prerequisite knowledge for Techno-Kleptocracy.

My argument in Techno-Kleptocracy is not about centralization per se, it’s that aggregators (by which I mean ‘companies who capture attention in order to sell it’) are able to take the money, efficiency and opportunities from everyone else.

I am not against ‘aggregation’ in and of itself. In the series of articles I wrote (referenced above) I am laying out the need for the aggregation of offers. What I’m saying is that aggregation is taking place in a way which benefits the aggregators at the expense of the public.

The only difference between other aggregators and DuckDuckGo is that DDG addresses privacy. It doesn’t address the underlying economic problems of homogenisation, profit extraction and so on.

Yes, filtering is required to make sense of the world. In ‘External Offer Processing’ I talk about maps which are simplified representations of our economy, society and information which we give our attention to (because they remove irrelevant detail).

The question of which detail is removed (and/or is irrelevant) is in the eye of the beholder. ‘Censorship’ is a filter imposed on someone who does not appreciate it.

I share your concern that exercising control takes time and energy - it will be the limiting factor for diversity in the system and the economy, and we will need to ensure that offerbots are not dominated by default options which everyone blindly uses (such that all we manage to do is distribute offer processing and not diversify it).

That said, it certainly seems possible to greatly increase the diversity of our offer processing, which is why I have persevered with this project.

My current assumption is that offerbots are quite different to existing tools and, therefore, we will need to build a lot of new things. That said, I’m keen to avoid reinventing things and would like to make use of existing technologies and communities where doing so will not adversely impact the project.

Hey @AndrewMackie, thanks for another iteration and providing your thoughts once more!

Thanks informing me of the updates on The Problem page and the new work in progress Purpose of Offerbots page – I have reread the former and will have to visit the latter once it is ready.

My comments on the Privacy aspect were not intended to be too dictatorial – whether it is something you think best addressed in the model, the problem, the solution, multiple places, or not at all. I merely think, based on your responses here, you’ve put in a good deal of work identifying gaps in other peoples’ work. Unfortunately the current state of the pages doesn’t exactly reflect that. I just wanted to bring that to your attention – otherwise you may tire of yet another person later down the line who mentions the same thing: “Have you considered X?” to which you’ve clearly demonstrated “yes”.

I think the key insight you just made – privacy/security first versus economics first – is really a telling difference between ourselves. I’d like to provide you a counter-argument, which will contain my opinion moreso than structural feedback, which you are free to use to sharpen your thoughts. Let me know if it was valuable, or a waste of time. Before that, I want to provide more feedback w/ as little of my opinion as possible.

I think your counter-points to Goldhaber would be valuable for your work to reflect. Again, up to you how to do so, I’m not trying to be dictatorial. But illustrating to readers how you’ve gone back to the late 90’s, updated the work, identified its flaws, and are providing a much more modern take is super duper awesome.

I do think I did get a little confused around the term “Aggregators”. I re-reread the articles around The Problem, so I finally think I am finally on the same page for the Techno-Kleptocracy bit. I understand your economic analysis and your view better now, maybe with the aid of any edits you’ve made to the pages in the meantime. Doing all I can to not interject my personal opinion, a suggestion: I believe it would be very helpful to include an argument for how, in the Offerbot model, there could not possibly exist any Aggregator that is helpful/not-harmful.

Additionally, would love for the work to include identified weaknesses (ex: your mentioning of “default options”) as that would also set the ideas forth on a road to strengthening.

Another suggestion is to explicitly mention the degree with which your economic model relies on distribution and decentralization. I think it is implied, but never expressedly mentioned. Which is a shame, because then you can analyze the market with the handy dollar-graphics like you did before. The economic analysis of Offerbots is currently missing, sadly! This will tie into the following counter-argument…

Warning, my opinions ahead

OK, here is the counter-argument I’ve developed:

My re-re-read of your documents makes me think you are really outlining three things:

  1. A model (it is part of “the-problem” pages)
  2. Identifying problems in the model (also a part of “the-problem” pages)
  3. Identifying solutions for the model

I think my initial response in this thread stemmed from the conflation of 1 and 2, which is why my gut-response included concerns about breaking the problem down and trying to tackle too much. It is really hard to independently view the model in a bigger context independent of the problem at hand, which makes it hard to play around with and bend because I’m probably going to violate a nuanced constraint baked into the model.

That being said, the model as I can best understand is something along the following (note: I would like to reproduce/butcher some of your images w/ modifications here, please let me know if you’d like me to remove them):

As I understand it, the key components in the model are:

  • Entities (users, etc) who pay attention and can generate an offer
  • Offers
  • One or more Offer Processors (only one universally shared one is shown in the above diagram)
  • Maps/Subsets/etc output by an Offer Processor (to which entities “pay attention” to)

My counter-argument is: focusing on the Privacy and Security aspects as a fundamental building block in a system ensures that the economic forces that come cannot allow people to, as you say, “voluntarily giving away their privacy and data to obtain offer processing from FAANG and other aggregators”. Whether that economic system is a laissez faire one, an authoritarian one, legalized or outlawed by nation states, those don’t matter. All cannot violate the Privacy and Security primitives. Consider your idealized outcome [citation]:

This requires individuals to afford:

  • The time and energy to set up an Offerbot
  • The money to pay for the computational power
  • The knowledge to run an Offerbot and understand its “proper” role in wider society

Real life is messy, and so individuals may naturally collaborate and share an Offerbot for various economic pressures. I understand this violates the verbatim “each person and organization will have their own offerbot” but currently the Offerbots documentation puts this statement down on paper but offers no description on why and how (your focus, I believe) the economic incentives will always enforce this. However, Privacy and Security primitives usually explicitly forbid this as a property via creative math. This insufficiency is reflected as:

You probably know where this is going next, given the economies of scale and various other existing influences today that push people away from self-hosting and towards leveraging shared computing power:

I think these pictures illustrate that the current description of Offerbots does a great job of using a broad economic problem as motivation for defining a model. However, it does highlight that a step is skipped: analyzing the model, breaking down the properties it has, acknowledging weaknesses, and therefore determining its soundness and viability for a solution.

Furthermore, the counter-argument is that the Security and Privacy properties being explored by existing projects (from Secure Scuttlebutt to pick-a-blockchain; from FreeNet to right-to-repair) typically do have as a core feature “sovereignty over their own offer processing”. I’m not sure if your paragraph meant to imply this was not the case. I think this is a shared concern, it’s just the privacy/security/economics are just (as you say) second-order motivations are ranked in different orders. So I am of the personal belief that the real strength in Offerbots is really providing an opportunity for existing solutions to determine what an economic impact could be and its resistance to re-centralization into Aggregators (see diagrams above). Thanks ahead of time: I know I will be using it as an analysis tool when I have the chance to look at what an “evolved ActivityPub” protocol looks like.

1 Like

Hey @cjs, thanks again for taking the time to read and offer your opinions. It’s very helpful and much appreciated.

My primary focus with the offerbots website has been to create a linear problem statement which will carry readers through the main argument. My original document was two hundred thousand words so there’s been a lot left on the cutting floor. Feedback such as yours, however, helps me to see which gaps are left and where I can either flesh things out in the main argument or create tangents for readers to explore. Some things can also become blog posts rather than part of the main thread, I suppose.

Yes, I struggled to know what to call them, but ‘Offer-Processing Corporations’ didn’t roll off the tongue so I took Ben Thompson’s lead and called them Aggregators. My apologies for the confusion.


To state the problem quickly here, the problems with profit-maximising aggregators are that they:

  1. sell attention by distorting markets (rather than sell maps by optimising markets),
  2. homogenise markets by centralization (reinvesting superprofits to dominate the race for capturing attention),
  3. use their control and efficiency to take vendors’ margins and efficiency, and
  4. replace vendors by selling competing products (and win because the vendors must compete with the aggregator on price while giving their margins to the aggregator).

You can easily have a benign aggregator (I am actively attempting to build one) by not maximising profit, allowing vendors who cannot afford to pay, minimising distortions and so on. The downside is that the aggregator is still homogenising the market.

The key benefit of offerbots is that you can create a market for anything / everything by simply creating an offer for it (or searching for offers) in a feedback loop. At the moment the only way that markets happen is that an entrepreneur has to notice an opportunity which is large enough for them to code and make a profit from. The economy is probably only a small fraction of what it could be because of this. As evidence, I cite the enormous popularity of the latest market for whatever new startup makes it possible (e.g. AirBnB).

That said, it’s important to note that these market makers are also providing a form of trust and, therefore, safety for people which will be a challenge for offerbots. The kinds of trust they provide are usually quite self-serving, however - often the aggregator will show only the ‘top’ results (by some arbitrary measure - popularity, previous ratings, etc.) which gives buyers safety at the cost of excluding the majority of vendors (and is yet another form of market distortion). I believe that we can leverage social networks to provide some or all of this trust without the algorithmic bias.

This is the purpose of the Challenges article, to point out the problems that I see with offerbots. It obviously needs more work to be completed (and it’s feedback which helps me to do that, thank you).

Yes, offerbots have a layer of decentralization to them (they are not merely distributed), in that we will be able to offer information goods, such as sets of offers, pre-processed offers and maps to each other. I expect that offerbots will have reputations for gathering reasonably complete markets for certain goods and that we will pay them for their data. They will not become monopolies, however, because the data can easily be aggregated by other offerbots (i…e they do not have exclusive relationships with vendors, they simply collate data).

This is important, because otherwise everyone must have an offerbot which is large enough (has enough bandwidth, storage and processing power) to interact with every other offerbot.

Incidentally, I see it as a strength that markets will not be complete in each offerbot (i.e. no offerbot will see every offer from every offerbot which matches their demand) because this also creates diversity (people will choose different offers).

Indeed. I shall have to correct that, but I’ll probably need help from other people to do that (because it’s harder to critique your own idea).

I love it! Butcher away.

Yes, aggregators create a two-sided marketplace (the ‘vendors’ who make offers and the ‘consumers’ who receive them).

Offerbots are intended to create peers, each of whom can make and receive offers of any kind. (FWIW Facebook also creates peers - every account can make offers of any kind, but it’s constrained by funneling everything through a single algorithm, excluding all of the offers which are not ‘engaging’, such that Facebook can charge those people money to appear in the news feed).

Yes and no. I assume that people will share offerbots and have considered schemes which would allow wealthier people to give virtual offerbots to strangers without compromising their privacy. We also need parents to be able to give their children offerbots with parental controls. And there’s problems with this when trust relationships break down (e.g. a married couple shares an offerbot but then they split up … how does each extract their data without one party abusing the other?).

I’m not attempting to mandate that everyone must set up and run their own offerbot instance, but I am saying that everyone must have access to offerbot functionality (which they alone control).

I’m also assuming that offerbot hosting will be a thing (just as WordPress hosting, say, is a thing), but this also creates uncertainty - how will the average consumer know that the offerbot host is running real offerbots (which protect their privacy) rather than fake offerbots (which steal their data)? I don’t have an answer for this yet.

As for the argument that the incentive for sharing offerbots will simply create another aggregator, I have already answered this (I hope) in that the offerbots will make the offers public, so aggregator of the offerbots and their offers gives nothing exclusive to the aggregator / host.

This is a good point and is worth responding to (in addition to what I said in my last point).

My assumption is that the aggregators will be so successful at taking everyone’s money, efficiency and opportunities that offerbots will be quite attractive to people who are acting as vendors. That is, they will be the only method by which you can sell your product without paying an aggregator through the nose.

Of course, aggregators have power over vendors because they have captured the attention of buyers, so the problem isn’t solved unless the buyers use offerbots too. Here I hope that there will be a few driving forces:

  1. buyers (‘consumers’) want to be vendors too (they are currently excluded from aggregators and they don’t want to be), and
  2. buyers will prefer the efficiency granted by offerbots over the convenience (efficiency without control) offered by aggregators,
  3. offerbots will give better results for buyers (except, perhaps, when searching for facts, which is the only thing that web search engines are designed to do and, therefore, are IMO the only thing that they’re actually good at), and
  4. offerbot owners will want their friends, family and others to have offerbots as well.

Perhaps, but AFAIK they are not attempting to offer generalised offer processing, which is what we need to crack the aggregator nut. It’s not enough to emulate an existing aggregator’s functionality, to extend it or even come up with something new. We need to give people the ability to have control over every offer they make and receive.

:star: Have a gold star for reading this far!

1 Like

Moved this topic to #fediversity:fediverse-futures and just created a brainstorming entry to Fediverse Futures on Lemmy:

Think some of the concepts behind this may be interesting to @bhaugen and @lynnfoster as it comes close to value flows…

1 Like

Certainly compatible. See Proposals · GitBook

Bonfire is implementing federated offers and requests.

1 Like