A community to talk about the Fediverse and all it's related services using ActivityPub (Mastodon, Lemmy, KBin, etc).
If you wanted to get help with moderating your own community then head over to !moderators@lemmy.world!
Learn more at these websites: Join The Fediverse Wiki, Fediverse.info, Wikipedia Page, The Federation Info (Stats), FediDB (Stats), Sub Rehab (Reddit Migration), Search Lemmy
Decentralized sales platforms would just suck to use, in general. The Amazon problem is likely something that can only be solved by the legislative processes of the countries it operates in.
Imagine Ebay but with even less scam prevention.
Wasn't a federated Amazon just a mall, in a way?
A mall is a private real estate instrument built by speculators to extract rent from businesses and it's actually rather predatory. This is fundamentally not real estate and fundamentally does not exist to extract rent, so it's more like "what if you took a mall and removed all the mall-ness from it".
If malls were collectively owned by the stores that comprise them and pieces of the mall could appear and disappear at will of whoever's participating... Is it actually even still a mall really???
Good point!
The mall was still centralized and most shops didnt have their own place and a stall ij the mall but I can totally see where you're coming from.
It might be a good idea to keep this in mind if this ever becomes reality and we need marketing ideas. :)
You don't seem to understand the retail operations of Amazon. They provide logistics and marketing services to retailers, they also directly compete against those retailers because those retailers can't do better at logistics and marketing without using Amazon's services.
I work in the IT department for a fairly large payment service provider. I can tell you now that you seem to be vastly underestimating both the financial aspect of this as well as several legal aspects.
Federation would almost certainly have to be opt-in rather than opt-out. I don't think you're going to pass KYC checks for any PSP if it's opt-out, the risk of someone (ever so briefly) selling illegal goods through your website is too great otherwise. Stripe would just shut down your account (if they even let you open it), PayPal probably won't let you open it at all.
Selling goods from other sites through your own, makes you liable for any returns, warranty claims etc... Simply "passing these on" isn't going to cut it. If the other site disagrees with the customer claim, you are on the hook for it, because it was sold through your website.
The financial logistics aspect here is really complex. If you're going to process payments on behalf of another site, you have to deal with reconciliation. After reconciliation you have to the send the money to the other shop, incurring additional (sometimes surprisingly sizeable) fees. And coming from someone who deals with (automated) reconciliation on a daily basis, every payment method does it differently and they all find extremely creative ways to mess up your systems. And that includes unannounced changes, mistakes, random unexplained fees, failure to deliver settlement files, etc...
How do you deal with the risk of scam instances? E.g. instance A tells instance B that a product was sold and the payment was processed. B sends it out, but it turns out the customer was the owner of A, and there was no payment at all. B just lost a product with very little chance of getting it back.
Then there's practical aspects. How do you deduplicate products in search? Or will you have dozens of listings for the exact same product?
The only remotely viable way I see this working is if only search is actually federated. Once you are on a product page, you can only pay using the payment page of the instance that has the product. You won't be able to pay for products of multiple instances at once, and you might lose some unified styling. But at least that approach has a chance of passing KYC and deals with all the legal issues regarding returns/warranties etc..., and it reduces the scam risk because you're in charge of your own payments. But at that point, you've only federated product search and nothing else, and then as a consumer you might as well just Google it instead.
I appreciate you have experience in running a business, but running a marketplace, especially a very complicated one, is really not like running a usual business.
This is incredibly valuable advice! Thank you so much!
My current stance on federation is of course opt in and requires the main seller to trust the downstream vendors.
The main point is that this already happens for a large portion of thing you can buy. I sell computers and adjacent services, classical system integration if you will. Of course I have to buy the systems from vendors and resell them to my customers.
Many system integrators have shops where some of them rely on custom integration of vendor apis. Take minecraft server sites for example that have an automated integration with a hosting company's api (eg hetzner). you as a customer just order a server, their automation makes the order processing with hetzner and provisions the server for you.
Now make this over a non custom but standardized api, eg activity pub.
I might still be overlooking stuff but from a technical standpoint this should be doable. The legal aspect is interesting, although I think this could be done similar to already existing resellers.
Feel free to point out flaws obvious to you. I appreciate your feedback massively.
I think the biggest issue is that if you already need to separate payments, returns, shipping, etc... you're left with a shop that also advertises products for other shops, possibly competitors. Then the question becomes... why bother federating at all?
I think it'd be better to set up a FOSS shopping platform, eg something that competes with WooCommerce or the likes. That's significantly easier from a financial and legal perspective, and I think it's an easier sell to actual merchants (why pay a license for that shit, use this one for freeee). Then once you have that running, you could think about optional federation as an addition to an already well-functioning platform.
I really don't see the appeal of activity-pub for this.
It's a protocol used for social media and interactions. You describe just sort of a "metastore".
Maybe a review store site could work better with activity pub.
Reading the post, I found what I really want right now: a federated review platform. Too many times I want to look for a product, and has to look into a reddit thread to see a recommendation. There should be one, right? Where is it?
Reading the post, I found what I really want right now: a federated review platform.
!neodb@lemmy.zip is a general review site. It currently covers media but, if you can get the data in (SKUs?) I can't see a reason it couldn't cover other products.
Thanks a lot! Bummer that it is primarily for media. I wish it had more publicity/popularity, is there any way I can help?
It's not a bunker per se, it is just that, if you are a review platform, the easiest initial targets would be Goodreads, IMDb, etc. Other types of stuff to review may take a little longer. However, if you can get access to a source of unique IDs, it may be possible to import the information. As it is written in Python there will be knowledgeable folks around who can better advise on this. I'd suggest post about it here: !neodb@lemmy.zip. See what other people think about it.
I think it makes sense. It would allow a decentralized unified search across all stores. With Lemmy I can search posts as long as the instance is federated. With this I could find products.
In this example instances are stores, stores are users in instances? How stores are protected to be defederated by competitors (we are talking about money and making a living here).
What it adds to just a simple centralized service that any store can join. If you don't want it to be another amazon, make that service a coop. or some kinds of non-profit that it's paid by the stores that want to become part of that.
I think here we are in the classic conundrum of "a solution in search of a problem".
Fediverse and ActivityPub is cool, but it's a social media thing. And decentralization is cool when needed, for instance social media. But it doesn't have to make sense for every use case.
For what's being proposed there's zero actual need for decentralization or ActivityPub.
Instances are stores (think Amazon or Etsy). Products are posts. Sellers are users.
Stores aren't protected from being defederated. You can still search Google or whatever, still visit the site and buy stuff. It just will not be a unified search, just like how anything else works with ActivityPub.
The good stores would be run by admins who don't have an incentive to defederate from others. Stores don't make money or take a cut from sellers anyway. The sellers aren't in charge of the instance, just like an Etsy seller can't do anything about the fact that they have competitors on Etsy.
The need for decentralization is that the store / Amazon / Etsy is broken up but the search and interactions, reviews, etc. are unified.
Those admins are unpaid?
Managing a store it's a LOT of work. And you are doing to provide profit for other people. Who is going to do it for free?
It's not like social media where people may volunteer to admin and mod, and users may donate because it's a common goal of share information, opinions, knowledge, funny stuff etc.
Here we are talking about bussiness that do what they do because they want money. I would not volunteer to admin a store so shop owners could earn money, that's for sure.
And I still not see the advantage of doing within the ActivityPub instead of just being a normal service where all interested shops could join.
What about the online food ordering market. I reckon that might be an easier first step than consumer products. Here in the Netherlands JustEatTakeaway has a market share of around 90% and requires restaurants to give them a 14% provision. Restaurants don't have much of a choice, if they're not on there they miss out on a huge part of the market, it's like they don't exist. Why don't restaurants unite and develop a FOSS protocol that let's them federate, so the consumer has a central place to browse the food delivery market, but simultaneously makes the providers independant because they can run their own instance if they please. Have these types of ideas been pitched to branche organizations? Restaurants have a clear interest to develop this to free themselves from the platforms with a monopolistic venture-capital-driven strategy.
I fully agree that this would be a valid application. The reason any company doesnt adopt such strategies is the cost of pioneering it. Most companies who spearhead such an idea want it to pay off -> proprietary. Also most people are specialized in their industry. Developing an app is not native to food industry for example.
I think the way to beat amazon is to specialize in one tiny area. Carve them up into such small slices that they cant fight back.
So like, instead of trying to do just their books business, do just horror books. Horror that mixes with all genres, every possible crossover, but always horror books.
Having a genuine specialty is what can take amazon down, bit by bit. Something genuinely cool, something genuinely fun. Another big-ass store is nothing special.
I would love a federated Amazon that works directly with producers to sell everything at cost without a middleman or fees to the sellers.
No, federated model is chosen over distributed model or centralized model to allow feuds, putting it simply.
That may work for a Reddit alternative, but doesn't work for markets. Helps moderation (some idea of it, I don't think that idea is good), but definitely hurts a single space to sell and buy stuff.
Which is why cryptobros and such types make either centralized or distributed systems.
So much for using computer networks for this.
Now about Amazon specifically - your post omits the whole warehouses and logistics part. Which is most of Amazon's core business.
Computer people today somehow started forgetting that real life is very hard and complex. When I was a kid (born 1996, so not old man), computers had a promise of making that real life easier, and from time to time delivered on it, but at some point bullshit like glossy buttons and Web 2.0 and social media became a thing in itself, and everyone started behaving as if it's done, we now can look down like olympic gods to those mortals messing around in dirt, and sometimes easily solve their problems. We can't.
Getting back to logistics - one has to design a system of shared warehouses, transportation, mailing and delivery tasks, tracking, reporting on outcomes of every event, and all that should be even more abuse-resilient than the processes inside actual Amazon. You'll have Byzantine problems in every interaction.
The "distributed king of all social media, solving once and for all the problem of centralized platforms" that I'm often dreaming about is realistic compared to that.
I know that Federation is exciting, but all these ideas for federated services are really missing the reason why the Fediverse's current bits are successful - because they have low moral hazard.
When you get into economics and meatspace relationships, moral hazard skyrockets.
It's much more than that. Amazon's strength is also in its proximity warehouses and contacts with delivery companies.
Otherwise you just have a federated Ebay.
you are not proposing a federated amazon, this is just federated ads and/or reviews.
how to process payments? how to ship goods? how to handle refunds? how to handle contestations?
please you can't just make anything federated. this protocol is built for social media and struggles to take over that sphere, we should focus on one thing rather than throwing random stuff at the wall hoping it sticks (cough federated tik tok cough)
how to process payments? how to ship goods? how to handle refunds? how to handle contestations?
The problems are solvable, but the solutions taken together are couple times as complex as Amazon itself. This translates to cost. Which is naturally the reason Amazon came to existence earlier than that solution.
I think that layers of storage\messages and actual logic should be firmly separated, an instance going down when someone wants a refund for an operation that involved it seems not good enough. If the operation is a cryptographic contract with an escrow, and "instances" are just servers providing message storage probably privileged for some users (might be members of a community, might pay for that storage, that's lower layer anyway), this is less of a problem. But that's not a federation.
By the way, however I dislike OP's attitude, if you suggest this idea like a federated ads and reviews platform, it becomes useful.
I don't see why we can't just buy directly from shops. Maybe an aggregator of links for products, so there is an rss-like feed of products, prices etc?
amazons true strength is ultimately in their logistics. Amazon itself isn’t a bad idea in theory but the execution is poor because of cutthroat capitalism exploiting workers and privatization. Ultimately the idea of sellers being able to ship their goods to communal warehouses for fulfillment should be a service that is nationalized. The marketplace can be federated, sure