Hacker News Reader

Comment by bastawhiz | Hacker News Reader

Neither did you. They release 3 datasets partially out of dozens

https://news.ycombinator.com/item?id=31385975

Your aggressive tone is completely inappropriate, especially considering that you are mostly wrong. They have released a single dataset, and specifically state on the site that only some of the data will be open sourced. So, perhaps it is you who should be reading further.

"Google shares some Crowdsource data via open-sourcing, for the benefit of the global research and developer communities. This includes the 400,000 images that we've already released as Open Images Extended, with more sharing to come."

https://crowdsource.google.com/about/how-it-works/

First, they've released three datasets, not one. It's wild to expect they'd release all of the data, both legally and ethically. Publishing data that turns out to be garbage is worse than publishing no data at all. Publishing unfiltered user generated content when they also have the resources to evaluate quality is silly.

They've built and maintained the site that nobody else is seemingly building and offering. They've already released data from the site that's high quality. They have committed to releasing more. What more do you want from them? All things considered, this is better than essentially every other big company is doing when it comes to giving away free training data.

The GP made sarcastic commentary about Google keeping and using this community-generated data to empower themselves. Your response was to belittle them and arrogantly claim that Google is releasing the data. You were then shown that Google is only releasing a small (as implied by "some") subset of data publicly, which makes the GP's claim mostly true.

But now for inexplicable reasons, instead of just apologizing to the GP for being rude, you're scrambling to apologize for a corporation whose entire existence is notoriously based around collecting, abusing and selling user data, while also brazenly colluding with all of the other big companies to accomplish that abuse.

And for the record, the "three" datasets you're attempting to nitpick -- again without doing any of your own research, but instead quoting some other comment -- are actually just three partial subsets of the same Open Images[1] dataset, which itself is one of many datasets. So, they've still only released (and intend to release) a small portion of the overall data.

Your only intent was to make someone on the internet feel stupid, when in reality they were mostly correct, and you are still mostly wrong. Your previous comment got flagged to death. Take the hint. Doubling down isn't productive.

[1] https://storage.googleapis.com/openimages/web/index.html

It's even wilder to say both

> they are releasing the data under a permissive license.

and

> It's wild to expect they'd release all of the data

and still have an attitude.

Reply on news.ycombinator.com