I also share your sentiment on this. Is Amazon currently restricted from using CC BY-NC licensed photos in ML training models because they lead to commercial benefit? If not, why not? If so, this change gives Amazon a much larger dataset to train on and is substantially beneficial to them.
Signing up to a program (particularly one Amazon designed) is much easier than negotiating a commercial contract and sale of data, which to me is what this is underneath. I agree with the sentiment of open data, but not with handing more of it to the ever more centralised holders of it, here being Amazon.
Has iNat checked that Amazon still provide the best value per/month?
Does iNat still retain all ownership of the data and its usage or does Amazon’s program give it additional permissive use (other than that provided by the users changing CC preference)?
I’d rather honestly that iNat just do a fundraiser and I’ll pay them directly to support the platform.
I’m personally happy for anyone to to use anything I put here non-commercially, but I have so many observations whose taxonomy is rapidly evolving / controversial that allowing them into the public domain would result in a lot of use that becomes outdated and incorrect rapidly (someone using my photo and calling it an ‘X’ on their site when it was an ‘X’ on iNat, but not updating it when the ID is corrected to ‘Y’)
I’ve had my default photo and audio licenses as CC-BY and my default observation license as CC0 for some time, to maximise use of my data. For anyone a bit confused about all the options, who does not earn money from their photos, and who wants initiatives like Wikipedia to be able to re-use media, those are good choices in my opinion.
Edit: I notice that the license options now show which ones will allow data to be shared with Wikimedia and GBIF - neat idea!
Gonna have to think about this one. While I am more than sympathetic to helping iNat out on costs etc, like others have commented above, I am a little gun shy having been burned before by finding my photos used for purposes I never would have approved of.
While I’m not so naive as to not understand there are people who will do that regardless of the license I apply, I’m not 100% comfortable with just giving a green light to it.
I much prefer what they’re doing now. Changing the license on my photos (which I had already done before seeing this) costs me nothing, and since I don’t mind if people use them there’s no reason not to. But the “please give us money” thing drives me nuts, especially since I couldn’t if I wanted to.
A couple of comments. I just set my settings to CC with attribution. Thanks for including the link to the settings because they’re not easy to find.
I don’t run across the iNat blog at all in my interactions and had no idea it existed. The post for this news is poorly configured for social media sharing as the featured image really doesn’t work to attract attention (it’s an image of the settings page). I suggest thinking about how blog posts will work when shared. Also, I didn’t see this post replicated on the iNat Facebook page or Twitter account where it could be easily shared. When it’s important to get the word out, it’s important to use all media effectively.
Huge corporations can easily afford to do small kindnesses to non-profits and they do gain something, but it’s not necessarily financial (directly). They get to point to the kindnesses they do – good public relations. Also, it can make some of the officers or owners feel good about something they do.
So I can believe that Amazon is doing something that benefits iNaturalist without costing iNaturalist and the observers anything, and I can believe that without loosing my cynicism about big corporations.
(My observations are already licensed as free for non-commercial use, I believe.)
I am generally in favor of sharing my photos (not that I think anyone would want them), but it does make me wonder.
If we all frequently give away our images, does that make it more difficult for professional photographers to make a living? (There are Creative Commons settings for Non-commercial use, I understand, but still).
FWIW, I cannot see Amazon doing this out of innate altruism or for the minimal good will such may generate.
it probably gets a tax write-off for a gift of services to a non-profit organization.
it gets good press
it probably gets the data structured in a way that is more useful for applications it may be interested in (most likely a really good data set for machine learning). this is not to say that others can’t benefit from this, too, but Amazon is in a particularly strong position to benefit if it wants to because it has huge resources to exploit the data.
it probably will be able to get information one way or another about who’s using the data. that could help them to discover potential recruits or potential competitors, perhaps giving Amazon special insight into new applications for the data.
people who want to use the metadata files must access through AWS (i think). (it doesn’t look like it downloads properly otherwise.) even though it’s free to use, you have to learn some basics about how AWS works, if you’re not already familiar. so it brings you one step into the AWS door, if you’re not already there.
that said, in today’s world, Amazon will find a way to win one way or another. so i don’t think it’s worth denying a win to iNaturalist in a futile attempt to deny a win to Amazon.
Regardless of hosting benefits, as soon as I discovered that I could set my photos to public domain, I did. I gain nothing from reserving rights to my crappy cell phone photos, and knowledge gains something. If someone wants to use them for their company Powerpoint, or as a basis for a multi-million dollar ad campaign, or wants to get it tattooed on their neck, go ahead.
If you are a professional wildlife photographer, I understand that you would want some protections on your work. For the rest of us, join the public domain. The world is better here.
(full disclaimer: I’m a strong advocate of CC0 for non copyrightable metadata)
I genuinely never quite understood the concerns of enabling license choice and potential usage restrictions for purely observational metadata. While I understand (and totally respect) the concerns folks may have about the reusability of their media, the idea of creating a patchwork of licenses for metadata directly undermines the ability of using iNat data for research with minimum legal barriers.
I actually applaud eBird’s decision (which is otherwise doing a terrible job on the media license front) when it comes to making all their observational data available as CC0. This decision maximizes the volume of research relying on eBird data, which in turns benefits the project (when it comes to applying for funding, establishing partnerships etc).
I think, but am not a site employee so this is not official one of the biggest issues is that the observation data AND text comments are covered by the same license application on a record.
Text can be copyrighted, to say nothing of the issue the site faces of cut and paste comments.
It could be fixed by having 3 separate licenses, but that gets even more unwieldy.
I’m the #1 anti-fan of Amazon, but I feel it’s important to realize that the issues here would essentially be the same if iNat was hosting observation images with Google, Go Daddy or Acme Web Storage Corp.
Each iNat observer chooses how they want to license their images. That license applies to any individual or company that might want to reuse the image, and the restrictions are the same regardless of whether the potential corporate user also happens to host the images or not.
If iNat observers choose to use a Creative Commons license, then their images can be available for various types of reuse (e.g. an image licensed using CC-BY can be used on Wikipedia). In theory, Amazon or any other digital behemoth can use some types of Creative Commons-licensed images for training regardless of whether they’re hosted on their own Amazon Web Services servers or Wikimedia’s servers, or anywhere else. The small amount I know about the corporate application of image recognition suggests unsurprisingly that it tends to be used in fields where people will pay for some service that is facilitated by accurate image identification.
In actuality, the range of datasets in the AODSP program (protein structures, genome sequences, satellite images) strikes me as being more compatible with the hypothesis that offering free hosting for big non-profit research datasets is a good PR move for AWS than that this is a cunning plan for Jeff Bezos to monetize citizen science. Remember, anything that is put on this platform is available to everyone, not just Amazon.
I guess it’s possible that Amazon could use iNat images to train a computer vision model how to ID 2,100 species of Euphorbia, but who would pay for that when iNat already makes that available for free?
The consideration here for iNat users should be “Am I interested in making my images available for public use?” If so, choose a relevant Creative Commons license and check the box to update the license on existing images. A side benefit is that you’ll be saving iNat money on hosting. It’s totally fine if some users (e.g. semi-professional photographers) decide they want to retain all rights. No one is forcing anyone to change their chosen license.
I agree, especially as it severely restricts the amount of data that gets fed to GBIF. I don’t think most observers understand it, so just end up selecting something that seems reasonable when they sign up. It would be better for basic observation metadata (date, location, community taxon) to be CC0, and have a customisable license for text comments.
Also, just came across this topical meme and thought many of you would appreciate it…
I don’t think the permissions are set properly yet for the AWS CLI access. @pleary may need to chime in here. But yes the official launch is on April 15th. I’m actually surprised you are able to do anything via AWS CLI given those permission errors I’m seeing. Out of curiosity have you tried https://github.com/inaturalist/inaturalist-open-data/tree/documentation/Metadata/Import/RDS to set this up remotely, because I believe that is working now but its certainly more involved than just downloading the bundle and setting it up locally. Again, pleary will have more information here.
We discussed issues with the size of the metadata a bit, including a smaller sample metadata bundle so people can play with it without downloading the whole metatadata. One problem with splitting it up is that its not clear to me that by the different CC licenses is the way most people would want it split. Other’s might want it split by geography or by taxonomy or by date, which is partially the beauty of letting people download all of the metadata to query the subset they want. The alternative RDS setup definitely takes more work to set up and costs a few cents an hour while the database is active. But it does get round having to download the metadata bundle locally and incur the storage, memory and CPU effects of having the DB run on your local machine.
We’ve decided to do exactly that for the reasons you’ve discussed. If its not in the version you’ve been playing with it will be in the launch version.
Another thing we decided to do exactly for the reasons you’ve discussed. Its not clear we can touch the registry page before launch, but we’re continuing to work on the documentation and will definitely get more info on licenses in that by launch.
Correct, we should update the API docs. That’s in there because we incur bandwidth charges when someone downloads images not in the new Amazon Open Data bucket. So that recommendation is meant to help avoid huge costs associated with people downloading huge volumes of photos. That will still apply to people downloading photos not in the open data bucket (unlicensed photos) but we will no longer get charged bandwidth costs when someone downloads from the open data bucket. It will be nice to no longer have to give this mixed messaging around using iNat data (e.g. please use our data for your research / please don’t download our data because we can’t afford it!) and instead just enthusiastically point researchers to this bucket and the associated metadata.
I think you’re spot-on with the tax write-off and PR theories. The other reasons aren’t impossible, but seem unlikely (not because Amazon has purer motives, but just because they’re unlikely to drive much revenue).
It’s a bad thing that Amazon is able to hide its profits so effectively, but we’re not really effectively combatting that by saying “I’m going to retain full rights to my images so that iNat gets less free hosting and Amazon can claim a smaller write-off.”
If folks object to Amazon’s monopolistic practices, or its malign effect on income distribution or the environment, or any of the many other dubious aspects of the company, it will be a whole lot more effective to campaign to get your legislators to tax and regulate Amazon.
This sounds like the perfect opportunity to plug an online event we’re organizing this weekend with folks active on iNaturalist and Wikimedia projects (@andrawaag@siobhanleachman@adzebill). If you want to learn more about the benefits of an open license (CC0, CC BY, CC BY SA) for your photos and CC0 for your observation metadata, and watch how we’re reusing content from iNat to feature on Wikipedia to illustrate biodiversity for millions of people all over the planet, consider tuning in: https://twitter.com/andrawaag/status/1374775607414915072
