Programs that respect your privacy

Mojeek Blog Post

Mojeek recently wrote a blog post about Privacy Browser Android. I thought it would be interesting to share some thoughts both about Mojeek and about how that post came to be published.

I have joked before that selecting a default search engine in Privacy Browser is like trying to find a good Defense Against the Dark Arts teacher. It has changed multiple times since I first released privacy browser. At one point I told myself that if I finished developing Privacy Browser Android and Privacy Browser PC and still hadn’t been able to find a good search engine I would, reluctantly, develop my own. I considered that the criteria for a good search engine would include the following.

  1. It must have its own index and not be depending on Google or Bing or anyone else for search results.
  2. It must not track users, but instead use only data from the search query to display ads.
  3. It must not track if a user clicks on an ad. That would be a huge change to the entire ad industry, but I think it would work based loosely on the principles Troy Hunt explained when he dropped all ad network code from his website and instead went to a sponsorship model.

The hardest part of building a search engine index from scratch would be the cost of building and maintaining an index that produced good results. Off the top of my head, I figured that it would likely take a billion dollar investment to build something competitive with Google before it would grow to the stage that it would be self funding. And trust me when I say that no part of me wanted any piece of that fight. All it looked like was years and years of hard work.

Enter Mojeek. So far they are executing according to all three of the requirements I listed above. They recently passed the five billion page mark in their index. It is still nowhere close to being competitive with Google for certain types of searches, like those in Cyrillic languages or some technical searches like C++ programming queries. But, it is getting better, and they will likely be fully competitive with Google long before I am fully done developing Privacy Browser. Remember, building a competitive index is a billion dollar experience, which is chump change for Google or Microsoft, but an unimaginably difficult hurdle for almost anyone else.

Another reason why I am glad Mojeek came along is because I think the general public is better served when the browser and the search engine is not developed by the same entity. Keeping them separate helps keep both of them honest. If, at any point, a browser stops protecting the privacy of the users, the search engine can write a new blog post and recommend a different browser. And if, at any point, a search engine stops protecting the privacy of their users, the browser can drop it from their list of included search engines or change the default in the list.

It is for this reason that I have committed that Privacy Browser will never monetize the default search engine. I have also publicly explained the criteria for including a search engine in the default list in Privacy Browser. That criteria may change over time to reflect changing realities in the search engine offerings, but the list will always be public and will always be applied fairly to all search engine. Privacy Browser will also always make it easy for users to select a custom search engine of their choice.

Because I feel so strongly about the need for a browser to never have a financial connection to the default search engine, and because I feel like full transparency on this issue is essential for users to being able to trust a browser, I would like to post relevant sections of email communications I had with Josh at Mojeek prior to their blog post going live. I thing the content of these emails reflects well on Mojeek and demonstrates their professionalism in corresponding with other developers. I sincerely hope that Mojeek continues to execute on the sound principles they have so far demonstrated and that they reach a point where their revenue is self-sustaining, a point I would like to reach as well ;).

April 25, 2022

Hello Soren,

I hope this email finds you well.

For a while now we’ve been intending to write something about PB; it is the only option out there aside from our app (which is nowhere near as advanced) which has Mojeek as default, and it would seem that PB is very well-liked by people we interact with, on community.mojeek.com and in 1-1 interviews, comments etc.

I’ve attached here something we put together off the back of me using it on my Android device for 3 weeks now, but we don’t want to chuck up an article unless you’ve seen what it says and are happy with the concept. I/we get that this is not a formal relationship, more a case of you making that decision due to the specific criteria you need to fulfill in order to be happy with a default search option, and we don’t want to posit it as such. We get that we could be switched out for something else (as and when you find a better option for this set of criteria and the people who are browsing with PB) but your browser ticks all the boxes for us when it comes to something to promote, and so I’m sending this to you now.

Let me know, feel free to wing specific edits if that’s the best way to go about things, or just say a flat no, which is all good with us.

All the best,

Josh

To this email I wrote the following response.

April 25, 2022

Josh,

I appreciate your email. You, of course, have no obligation to reach out to me before publishing whatever you like on your own blog, but the fact that you did so as a courtesy is reflective of the degree of professionalism I have come to appreciate in Mojeek.

I think the blog post itself is well written and I have no comments or objections to it being published as it currently stands.

Although it isn’t necessarily something you would put into this blog post, it would probably be interesting for Mojeek to know that most of my current effort is being put into developing Privacy Browser PC. Although the timetable isn’t set in stone, I would like to have an alpha version available with Debian packages by the end of the year, with plans to support both Linux and Windows in the general release. Mojeek is currently the default search engine for Privacy Browser PC as well, and I do not currently see anything on the horizon that would change that.
I intend to make Privacy Browser PC free on all platforms, with the development funded through donations. I also expect there to be a halo effect, with some users who find it on the desktop also purchasing the Android version from the Play store.

I recently posted a little bit of information about Privacy Browser PC on Mastodon.

Soren

A couple of days later I received this response.

April 27, 2022

Soren,

I’m very happy to hear that; especially when it comes to smaller teams, independent developers etc. our approach is to always make sure they’re at least aware ahead of time. Also thanks for the well written.

I think there might be a minor change to the title, something like “Which Browser Should I Use on Android?” and I’d also like to ask if you’re comfortable with us using the PB icon/logo as the header image, on a plain background (likely HEX C7C7C7 or 1E1E1E so that the logo stands out). The other images in the piece are just snaps from me using PB, so nothing that you wouldn’t have seen before, but I’m happy to pass them across if you have any interest in that.

The Privacy Browser PC is for sure interesting to us. Bare minimum I’ve boosted that toot, and fixed the egregious mistake that is not following your Mastodon from the Mojeek account.

Josh

To which I replied.

April 27, 2022

That would be fine. My understanding of copyright law is that using someone else’s logo under those circumstances (a review of their product) is considered fair use, so you don’t need permission. But once again, it is kind of you to ask first.

Soren

I had not previously had any interactions with Josh, but I found this nice post on their page explaining a little about him.

Last updated

,

4 responses to “Mojeek Blog Post”

  1. It’s weird to see a search engine wishlist which is essentially “clone Google and their business model but slightly better”. Why allow advertising at all? No other layer of your software stack is injecting ads into user content — even remote services, or layers which require ongoing maintenance.

    You’re not pushing a clone of Netscape Gold for the web browser, so clearly you realize that “paid proprietary product” isn’t the only way to deliver software.

    • I don’t mind looking at ads (see the reference in the above post to the Troy Hunt post on the matter). However, I find it completely inappropriate when an ad starts looking back at me through any sort of tracking. I think this is significant, and is why what I am proposing is very different than just being “Google and their business model but slightly better”. Google is all about tracking users. What I propose is all about not tracking users, but having appropriate ads based on the context of the search URL. So, if a user searches for plumbers in London, there are a couple of ad slots at the top that plumbers in London can pay to show up in. But no tracking of users has to be involved to make that happen.

      Running enough servers and consuming enough bandwidth to index the web is a very expensive proposition. If it isn’t funded by ad revenue, how would you propose it be funded?

  2. One thing I didn’t post above in my list of what an ideal search engine would do, and one of the core things I would do if I ever had to develop a search engine, would be to post all the code for the entire frontend and backend and package it so that anyone else could easily spin up a competitor using the same code that I was using.

    Among other things, this would keep me honest, because if I ever started doing something abusive it would lower the cost for someone coming in to replace me (at least the development cost, which isn’t really the expensive part of producing a full index of the web). As a byproduct, it would also allow people to use the system in interesting ways, like creating an index of an internal network.

    One key aspect of this would be that the ranking logic would be entirely public. There would be no secrets about why one result ranked higher than another. And all discussion about what this logic should be would happen in a public setting.

    As far as I know, this is not something that Mojeek currently does. But I think their offering would be significantly improved if they did.