Read on Twitter
What is the time for Google? As we speak, this engineering problem leads to a lot of internal discussions at Google. Let me explain what happened in the next few tweets. It all started with a bug report about Google Images. Here we go. (1/.....)
So the bug report is filed on behalf of a weapon expert who tries to find out, amongst other things, which explosive devices are used in incidents, the brands and introduction dates, and all kinds of other details that will make some #osint people burst into tears of joy (2/...)
Anyway, the weapon expert who asked for my help, noticed that search results in Google Images were suddenly not as good as they were. He is using keywords and then sorted on date. A Google engineer filed a bug report on our behalf. (3/...)
The bug generated a lot of internal discussions, as we speak, at Google. I have confirmation now that it is a real bug and not a feature. Oddly enough, the query we did was so specific, that it was never tested by Google. (4/...)
But it's getting addressed, and it's much more interesting than you might imagine. It touches on issues of "what is time on the internet anyway?" (5/...)
What is the time on the internet when we talk about geolocation and verification? Do we seek for the exact moment the photo has been taken, so the photo-taken time? If not, are we interested in the moment that the material was uploaded somewhere? (6/...)
Or should Google just look at the crawling date? That is jargon for the first time they see it. Wait, there is even another option. In documents, should Google monitor the first-placed-in-document time? (7/....)
This gets mind-boggling complicated when you want to research composite images on news sites. Which image, out of a set of 5, is the actual image time? (8/...)
In other words: what is time for Google. That is now the debate in Mountain View. Personally, I would go for crawling time, the moment Google sees something for the very first moment. (9/...)
But if Google decides to do otherwise, I hope they stick to their choice and will not mix the different views on time. Otherwise, it will be harder for fact-checkers like @factchecknet to find a primary source (Patient 0). I will keep you posted on this. (10/10)
