I would argue that a search engine can't be truly privacy focused without independence. Mojeek is one of the only ones that falls into this category. Compare their policy with Neeva's - https://www.mojeek.com/about/privacy/
Ideally, the HTTP/proxy layer of the internet could have solved this with caching of the search URLs, but caching has pretty much been gimped thanks to "privacy", HTTPS, logins, etc. I'll add this as another data-point to my tin-foil hat explanation of why privacy and HTTPs has so heavily been forced upon us by Google et al.
First, thank you for this feedback and everyone else's in the thread, as well. We are listening and improving based on what we hear from our early users and all those interested in our work.
For context on some of the choices we made when thinking about building a private search engine:
- First, one of the biggest risks to people’s privacy online is advertising. To present a true alternative requires an alternative source of revenue than advertisers. By being a subscription service we only put users first.
- Second, a search engine that stores nothing can’t get better. Even on this thread people have preferences like wanting to never see Pinterest results or see more of them. We want to offer a better search experience and privacy so you don’t have to choose one or the other.
Unfortunately, we can’t build all things ourselves to get there. Whenever we work with a partner, we are very careful to ensure that they see only as little as they need to see. For example, while we do use Stripe to process credit card payments, they are never going to see what anyone searched on Neeva. However, in order to cover the full, diverse spectrum of what people search we have partners like Yelp, Xignite and Bing in addition to our own tech. Given what’s required, there isn’t a viable, private search engine that exists without such partners, as well. Whether others disclose them is another question. Particularly the advertising partners of the “free” search engines...
In each of these cases, we vet the privacy standards of these businesses and make sure they align with our privacy policy and values, as well. In cases like Bing we take proactive measures to protect user privacy by obfuscating or omitting all other personal information outside of a query.
Therefore, clauses like the above are meant to help bring transparency more than anything.
Lastly, in terms of aggregated and de-identified information, we’ll take another look at this language. We've already updated our policy based on user feedback and we have no intention of ever sharing de-identified information like the above. Thanks for the callout.
Keep the feedback coming! We're listening. This is hard stuff and we appreciate all the support we receive.
https://neeva.co/privacy
They still plan to collect a lot of data, and share it with:
Microsoft is one such provider where we may share limited personal information in order to make the Services available. You can read about their use of data in Microsoft's privacy policy.So, Microsoft's frequently-changing (last updated: January 2021) privacy policy effectively applies:
https://privacy.microsoft.com/en-us/privacystatement
While more competition is always good news, I can't see this as a privacy-conscious choice.
Edit: You can even connect your Google account, and they will collect the data about you and your contacts from it. I wouldn't be surprised if they end up getting blocked for this by Google on, ironically, privacy grounds (Google has no interest in assisting an emerging competitor like this if they can get away with it, and they probably can).
Then, there's this gem:
We may aggregate or de-identify the information we collect. Aggregated or de-identified data is not subject to this Privacy Policy.
It's been demonstrated before that this is also a privacy risk. AOL's "de-identified" release of search-engine data turned out to be a privacy disaster:
https://www.cnet.com/news/aols-disturbing-glimpse-into-users...
https://www.cnet.com/news/aol-apologizes-for-release-of-user...
A privacy-respecting search engine should just not collect all this data in the first place.