This is precisely what a company called Sold used to do before their acquisition:

http://techcrunch.com/2013/11/04/dropbox-snatches-up-sold-th...

It's an interesting system to build, though, and there is money to be made. The very first version of my startup was something that was pretty much the same system, but we focused on offloading liquidation inventory. Some things that you will likely discover along the path

1. If you use eBay as your dataset, you are pretty much SOL. Their lack of rigorous moderation in terms of their image style guide makes computer vision really, really difficult to carry out. Using SIFT image features from the same exact images, I only got about a ~75% accuracy because eBay users tend to slap all types of random extra things onto their images. I can't imagine you'd have much success trying to use CV for identifying (with sufficient accuracy) consumer electronics in consumer supplied photos. I can't speak for Mechanical Turk... We wound up modifying our tech strategy such that we didn't need high accuracy in identifying other listings.

2. Product versions dramatically alter product value, but are just about impossible to figure out without being explicitly provided that data by the person holding the product.

3. Unless you own your own sales channel, you will be beholden to the marketplaces to offload the inventory in. In my case, we automated the process of selling on both eBay and Amazon. We were reading in data from multiple wholesalers (dropshippers), which had catalogs full of an arbitrary assortment of products... Which means we wound up listing weapons, sexual objects, etc. We got dinged pretty bad by eBay for these things (though, turns out batons sell really well on eBay). Eventually we pretty much got banned from both marketplaces for listing violations + bad fulfillment speed. We couldn't control the shipping speed of our dropshippers (we had no bargaining power), which I attribute to killing that section of the business.

As somebody who actually built the same system that you're describing (sans being an open marketplace, which actually makes it more difficult), I can say that you're in for a rough fight. Pretty epic watching an ebay store run itself, though...

Already done:

https://news.ycombinator.com/item?id=6660089

http://i.imgur.com/mmPAztu.png

I didn't get much interest so I didn't bother building it :)

I built a prototype too: thingsy.co

It finds all the pictures on Instagram with the hashtag #forsale, post them on Craigslist, and contact the seller through Instagram when a buyer emails the Craigslist contact info.

Craigslist killed it.

there is a lot of activity in this sector right now, including some VC funded startups, the approach they are taking is to grab instagram #forsale pics then post it on their own storefront site

Really? Do you have any example?

this is the best one I seen - https://10s.ec

a few others

http://new.soldsie.com/instagram-selling/

https://paytagz.com (Member of http://boost.vc)

Cool thanks!

It is a nice idea, but in your two simple cases the photo would be of a "ubiquitous" Xbox 360 and a iPhone 4S. But it could be of an iPhone 4 and in the case of the Xbox the date of manufacture really matters. The older model that _will_ one day die on the owner are worth less. And then there isn't any mention of how much space you have on them. But above all a photo might not indicate if the device actually works...

Would you be mad if your barely used unlocked 64GB iPhone 4S was sold as a non-working locked 8GB iPhone 4 and they send you a check for $10?

Most people don't know much details about things they own. They don't know the model of their router, they don't know the CPU in their laptop, they don't know the version number of their phone OS. What do they do? They usually turn the thing upside down (or go to the "device info" section) and find the data there. Snapping a picture or the sticker on your fridge/dvd/lawnmower is probably quicker than writing it by hand.

If the user can see these informations, he can take a picture of it. If he can't, well most people probably wouldn't know either.

Also, nothing stops you from taking a short video of the item and saying additional information out loud.

I love this idea because I think this is a huge win-win for both buyers, sellers, and efficiency (more stuff gets used, less gets tossed).

As the comments have pointed out the difficulty is accurately identifying items.

With used books you don't have that problem because every book has an ISBN number so you can easily identify any book. That's enabled a robust used book market.

So maybe the business isn't the end product, maybe the business is an intermediate step, the equivalent to ISBN numbers for everything.

The business I'm thinking of is QR tags for everything. The QR code allows the owner to simply scan the QR code when they want to sell their stuff.

You'd have to work with OEMs. The pitch to the OEMs is that a QR code that allows the item to be identified and easily resold enhances the resale value, thus enhancing the initial value of the item.

Yes. I quickly realised that the business is indeed the item identification mechanism. I used to call the idea "Facebook for Things", where everything has a profile.

Part of the idea was to identify stuff, and then send people a QR code so that they could refer to physical item with a simple scan.

Want to find someone that can repair that particular guitar? Scan it, pick "Find repairman" (or any other contextual action), and you're set.

Insurance backlog, selling, lending, sharing, hands-on trial (before buying online), lost & found, quick review, ownership history, package tracking, user's manual, community hacks and mods, 3D print a case/accessory, buy related accessories, brag, home automation, etc. Once you can refer to a specific item precisely (knowing everything about it), the sky's the limit.

I looked at existing barcodes you can find on some items, but they're rather limited. Recognizing games, music, movies and books is easy. But that's quickly becoming all digital.

I like the idea, but would love to see it taken a couple steps further. To me, the biggest issue with selling things online is dealing with the people who are interested in purchasing them. I have had less than stellar experiences with both eBay and Craigslist. I would love something like an expanded Gazelle that was not just electronics. Perhaps even with physical locations for things like furniture, exercise equipment etc. The company could then either sell on consignment, charge storage, or more riskily buy and try to sell at a higher price point. I have a spare bedroom filled with crap I would love to be rid of but dread dealing with.

I too have had less than stellar experiences on both eBay and Craigslist. The last time I tried selling on Craigslist I listed a brand new lamp, unopened in box. I included all the manufacturer's pictures and details (dimensions, weight, ect...). I received an email the next day from a lady interesting in purchasing it. She asked if we could make the exchange in a Walmart parking lot (in retrospect, that should have a warning sign). 5 minutes after the agreed upon meeting time I received a text that she was running late but would be there any minute. 15 minutes later she arrived and instead of apologizing for wasting my time by running late she insisted on opening the box before purchasing it. I refused since I was worried this would lower the value if she chose not to purchase it. She was being stubborn though so I eventually acquiesced. At this point I really just wanted to offload the lamp and go home. Upon inspection she said that she was sorry but wouldn't be buying it because it was smaller than she thought it would be, never mind the fact that I included the dimensions in the listing. As I was about to drive away, she offered me half of the agreed upon price. Why she suddenly wanted it again despite it being too small I'll never know as I immediately sped away.

Like you, I have a lot of stuff I'd like to sell. What I want: a company to send me pre-labeled and pre-paid shipping containers that I then fill with my stuff and leave on my front porch (safe, rural neighborhood) to be picked up. Every week or so they pay me until everything is sold - minus a commission so that our interests are aligned to get as much as possible for each item. I can't seem to find a service that makes it that easy though.

Oh, and that lamp? Amusingly, I took the lamp back to Home Depot where I had bought it new 3 years earlier. I had no receipt and I told them honestly how long ago it was originally purchased. None the less they offered me a full store credit. The catch was that they no longer carried that brand of lamp and hadn't for over 2 years. In fact, it had been so long since they last sold the lamp that they didn't have pricing information on it in their database. Since they couldn't determine how much I paid, they instead gave me store credit equal to that of a lamp they currently had in stock that we agreed was comparable. I was so impressed by their generous return policy that I promised to one day spread the word of how awesome they are.

sounds similar to offerup (http://www.offerupnow.com)

there is a startup doing this and launching soon, can't track down the link of their site right now