Hacker News Board Hacker News, browsed like a classic message board
Live data from Hacker News

Show HN: I'm building an open-source Amazon

openship.org

Post reply on HN
Print view View original

141–150 of 309 posts

Re: Show HN: I'm building an open-source Amazon

#141

Post by johannes1234321 »

swyx wrote: post #32
jpatt wrote: post #24

A few points I'm skeptical of: simply connect to the seller’s infra instead of locking them in. One of the value-adds of Amazon is there's a whole bunch of tech sellers don't need to know how to write, run, and maintain. You'd need to understand how big the niche "technically literate to run their own e-commerce business, but selling on Amazon for other reasons" is, then how your operation would answer those "other r…

i think the branding "open source Amazon" is just way too ambitious/big. invites a lot of confusion/criticism if Amazon is different things to different people.

Amazon to most is primarily a website where you get everything mostly reliable and mostly with a consumer focussed service (I for one never had any issues when returning things, any problem was like "yeah ok, send it back, we send you a new one or do you want a refund?" ... For sellers they play a different game with their market power, discounts or they won't offer your things or rank competitors higher)

Re: Show HN: I'm building an open-source Amazon

#142

Post by hk__2 »

rocket_surgeron wrote: post #44

Those efforts look impressive but no service will be a version (open source or otherwise) of Amazon until it can place a single tube of toothpaste in my garage 18 hours after I ordered it, which it just did thirty minutes ago. Amazon’s “tech” stack, both consumer- and seller-facing is horrible and borderline irrelevant. Their fleet of trucks and aircraft is not.

I’m pretty happy to live in world where it’s difficult to build a business where you move trucks around cities to deliver a single tube of toothpaste to someone who ordered it the day before. Yes, it’s hard to do this as fast as Amazon does. But should we really go that way, anyway?

Re: Show HN: I'm building an open-source Amazon

#143

Post by kornhole »

jpatt wrote: post #24

A few points I'm skeptical of: simply connect to the seller’s infra instead of locking them in. One of the value-adds of Amazon is there's a whole bunch of tech sellers don't need to know how to write, run, and maintain. You'd need to understand how big the niche "technically literate to run their own e-commerce business, but selling on Amazon for other reasons" is, then how your operation would answer those "other r…

>One of the value-adds of Amazon is there's a whole bunch of tech sellers don't need to know how to write, run, and maintain. You'd need to understand how big the niche "technically literate to run their own e-commerce business, but selling on Amazon for other reasons" is, then how your operation would answer those "other reasons" to make you competitive to the niche.The bigger stores with IT staff and budgets would be first to join and contribute. Some of us consultants could help the smaller shops and possibly combine energies into communities or coops using shared resources.

Re: Show HN: I'm building an open-source Amazon

#144

Post by ganeshkrishnan »

We already do this via hypernile.com although we are not open source.

The biggest challenge for you will be shipping from the 3pl. Amazon warehouses all talk to each other and it helps them keep the shipping costs bare minimum.

We use machine learning at scale to replicate it and we have made good progress however we haven't been able to distribute inventory between warehouses. Amazon can do it because they have their own distribution trucks while we are surviving on eating ramen for dinner

The hardest part is convincing warehouses to connect into our network

Re: Show HN: I'm building an open-source Amazon

#145

Post by tzs »

ryan29 wrote: post #46
eastbayjake wrote: post #30

Love open source projects in the eCommerce domain, especially ones that are JavaScript instead of PHP! Two pieces of feedback: - Using a copyleft license like AGPL makes this an automatic non-starter for most businesses, no matter how impressive your tech might be. You'll have a lot more luck with mid-size and enterprise adoption with an MIT license. - You've really built an Order Management System for marketplace us…

> - Using a copyleft license like AGPL makes this an automatic non-starter for most businesses, no matter how impressive your tech might be. You'll have a lot more luck with mid-size and enterprise adoption with an MIT license. What makes the AGPL unattractive? I thought it was basically just the GPL with a limitation on using the software to provide a SaaS product. You don't even have to contribute unpublished chang…

> What makes the AGPL unattractive? I thought it was basically just the GPL with a limitation on using the software to provide a SaaS product. You don't even have to contribute unpublished changes, right?

The things AGPL adds to GPL don't just affect people trying to do a SaaS offering of the program. If you modify it and users interact with it over a computer network you have to make source for your modified version available to them.

For example suppose it was software to add online ordering to restaurants. A restaurant modifies its copy so that it can be given the recipes of the items they sell and the modified software uses that information to allow customers to easily exclude items they might be allergic to or that violate their religious or ethical eating rules.

If that restaurant wants to use that as a competitive advantage over other restaurants they aren't going to want to have to give away their modifications, so aren't going to want to use AGPL software. They'd probably be fine with GPL software.

Re: Show HN: I'm building an open-source Amazon

#146

Post by hk__2 »

desiarnezjr wrote: post #136
the_cat_kittles wrote: post #118

Earlier quoted context omitted.

getting stuff that fast is very stupid 98% of the time. people may like it and want it and sing its praises, but its mostly stupid. and i think stupid things generally tend to die out. that is all to say that maybe that edge that amazon has is not that important.

It's not stupid. Unnecessary most of the time but it becomes an expected processing and delivery standard, which makes the experience much more predictable for consumers. Very few can really compete effectively with Amazon on this front. Source: Used to run an e-commerce company that shipped up to 2 million packages annually. Once we optimised shipping and processing, customer service inquiries and complaints dropped…

> Source: Used to run an e-commerce company that shipped up to 2 million packages annually. Once we optimised shipping and processing, customer service inquiries and complaints dropped dramatically, and customer trust / sentiment skyrocketed.

I used to work at company that shipped ~30k packages a day. I’d say 90% of customer complaints were about the delivery delays, and only 10% about the rest. It was really hard to optimize because we had 1M products and were using just-in-time logistics with very little inventory. This is a specific market that Amazon hasn’t really entered yet, but when they’ll do they’ll crush everyone.

Re: Show HN: I'm building an open-source Amazon

#147

Post by hk__2 »

throwayyy479087 wrote: post #49
rocket_surgeron wrote: post #44

Those efforts look impressive but no service will be a version (open source or otherwise) of Amazon until it can place a single tube of toothpaste in my garage 18 hours after I ordered it, which it just did thirty minutes ago. Amazon’s “tech” stack, both consumer- and seller-facing is horrible and borderline irrelevant. Their fleet of trucks and aircraft is not.

Amazon.com is likely the most a/b tested piece of software on the planet. It is ugly and clunky because that's what _works_ - same with Alibaba and Yahoo JP.

> Amazon.com is likely the most a/b tested piece of software on the planet. It is ugly and clunky because that's what _works_ - same with Alibaba and Yahoo JP.

This was on the HN frontpage yesterday: "Be good-argument-driven, not data-driven" https://news.ycombinator.com/item?id=32651763

I’ve seen countless bad decisions being taken because the A/B test was badly done. You can’t assume that some decision was good JUST because they A/B-tested it.

Re: Show HN: I'm building an open-source Amazon

#149

Post by swalsh »

PaulDavisThe1st wrote: post #139
swalsh wrote: post #125

Earlier quoted context omitted.

lol, do you end every technical argument with "I'm Paul Davis, don't even try to debate me!"

I try not to. But when someone claims that building a platform for distributed, cooperative, coordinated action is "trivial", and it's in the context of "I'm building an open source Amazon", it is a little hard not to fail at this goal.

I used to work in large scale eCommerce too, I would never claim distributed supply chains is trivial. But this was not in reply open source Amazon, this was a reply to the idea of a distributed Uber (the parent comment to mine).

That to me, feels quite in line with Web3's capabilities. The hard problems to me don't seem to be technical in nature. I think the bigger hurdles will be business related. But this is a tech conversation, so here is my thoughts.

The way I see it, Uber enables 3 primary capabilities, for which it takes a major cut, and commands total control over drivers:

1. Driver Ratings

2. Payments

3. Match Making

I think web3 can solve 1 and 2 on-chain using smart contracts relatively simply. and if you use a chain like Avalanche, it can be done with low fees (cents) and with fast transaction finality (seconds).

The match making capability should not be on chain. For that i'd design a simple rest API using traditional technologies that consumes configuration from the chain. This service can be deployed on Akash, and paid for by taking a small cut off each transaction or devaluing a utility token... i'd build a DAO to govern the whole thing, and i'd distributes votes based on activity in the drivers pool.

Are there probably problems here? yeah, this is 3 minutes of thought. But i'm sure if I cared to, I could design a functional system in a weekend.

Post reply on HN