Comment by kodablah | Hacker News Reader

Comment by kodablah | Hacker News Reader

I'm confused about a couple of things. First, the tech. So my browsing history is sent over the internet to a third party who comes up with recommendations via some magic opaque system? Can some recommendations be prioritized by how much you pay this third party? I have all sorts of questions about Laserlike's use of the data, but I guess I can dig for myself. Sure, you can say "well, don't use it" but this is more about the health of internet users in general, not just me.

Second, the reasoning. Why? Is Mozilla altruistically trying to help people find new things similar to what they're looking at on the internet? Is this really a problem begging to be solved? What is the relationship with this third party company Laserlike? Which way is money changing hands and how? How do resources get allocated to projects like this as opposed to any other (i.e. curious about prioritization)? From the outside looking in, it seems to have little value, but of course that's why you test these things. It just seems like a strange direction to put resources in.

I personally hope this experiment fails and people do not use their browsers for webpage-like things such as recommendations. This is not the browser's role and I'm becoming increasingly concerned at what larger companies think the browser's role is vs little ole me. My only reason for not wanting it to fail is I know that people put work into it.

Nick Nguyen's post on Context Graph¹ touches on the motivation, but in short, we think the health of the Web is threatened by existing players building experiences that, like a casino, you can link into easily, but which are deliberately difficult to navigate away from. For example, Instagram has historically restricted² which users are allowed to include external links in their posts. That harms the Open Web and concentrates power over what you can see and discover online.

On the other hand, if the browser itself can offer links that break out of those walls, then we can sidestep the existing filter bubbles and make the Web a more competitive, plural medium.

Per the announcement post³, if this proves useful, we'll look at alternative means of generating the recommendations; the current implementation (sharing history) is an opt-in shortcut while we figure out if recommendations are worth pursuing at all.

While we are searching for privacy-respecting ways that we can diversify Firefox's revenue, and recommendations could play into that, that's not what this experiment is testing. Further, we will not pursue this outside of Test Pilot if we can't do it in a way that's private and which adds genuine end-user value.

¹: https://medium.com/firefox-context-graph/context-graph-its-t...

²: https://techcrunch.com/2016/11/10/instagram-will-now-let-cre...

³: https://blog.mozilla.org/blog/2018/08/07/firefox-offers-reco...

Thanks for the response. I think by your phrasing that this experiment is not testing revenue generation you are also saying there is no revenue relation at all, correct? (well, maybe for Laserlike, heh) Disclosure of monetization of data in any form IMO should be a requirement, so I hope that its lack of disclosure means it is not occurring.

While I agree in general that alternative web traversal techniques could be explored, I am not sure I want my browser vendor + third party doing it. I understand the places we put in our address bar may have restrictive rules, and it's why I don't enter those places into my address bar. In good faith, if this experiment is deemed successful enough to improve/continue, I would ask that the role that Laserlike plays be pluggable with an open protocol. I would also ask that if it's adopted more mainstream that it always remain opt-in.

I know that it sometimes seems unfair to ask for all these freedom-respecting things while the other guys get away with so much. But what seems like a double standard is what endears trust.

My prediction if the experiment is a success would be: Mozilla acquiring Laserlike and trying to find out how to implement their technology client-side as best as possible.

Aren't deep hooks into the browser just making it worse? There are billions of websites and myriad search and sharing sites, and only a few browsers.

It just seems like something that can happen client side. Say there are deemed 100 new interesting articles in a day, send those 100 links to the client, and let the client rank them.

Thats how the recommendation by pocket works, right? Whenever I see these types of systems, I worry they try to hard to show us things more like our past, and dont try hard enough to shape our future. A recommendation engine should be more focused on "this is interesting, this is good for you to learn about" in a healthy way, compared to "this is more of the same but we know itll keep your attention better than facebook."

These AI recommendation engines ALWAYS fall too close to clickbait, like a drug dealer, and never close to a medical healer, currated lists like techmeme/mediagazer/longform/longreads/redef/aldaily. Prismatic and Trove come to mind. I bet one could get better results just parsing those 6 sites for new daily links.

http://www.bbc.co.uk/blogs/adamcurtis/entries/78691781-c9b7-...

> I worry they try to hard to show us things more like our past, and dont try hard enough to shape our future.

I really don't want some third party trying to "shape my future" by influencing my browsing habits.

I do. How do you get your news now? Google, facebook, some kind of feed reader? This site youre on? Youre already being influenced.

What about the education system? You wouldnt go back to school if it was free?

The important part is that its either opt in, or very clear to opt out, and that I can choose the organization I get my, for lack of a better word, education from.

The comments always look the same when Mozilla announces something like this. Perhaps the content team should have a specific checklist they run through when announcing third-party collaborations, answering questions like:

- is anyone making money off of this?

- what is the motivation for the third party to take part?

- is this the default, and if not, what extra demands would it have to fulfil to ever become default?

- what user actions, if any, can trigger interaction with this third party?

I'm sorry, and I respect your work, but I think this is nonsense. Let's say Mozilla moves forward with this project. Assuming that it will be a source of revenue at all, the "recommendations" are going to be based on whoever has the money to pay Mozilla for ad placement (and let's be honest that it is ad placement; if it's keyed to the specific interests of each user that just makes it targeted ad placement). That results in the opposite of the supposed goal of decentralizing power online. To be honest I'm not convinced that adding "genuine end-user value" is not just a paper-thin justification for accepting new sources of revenue that are antithetical to Mozilla's historical values. [1]

But let's assume the most charitable and best case scenario, in which Mozilla continues the program but doesn't accept any money for payed spots or higher placement. How are they going to automatically determine what is valuable? If the algorithms go by what is popular, then it's just a new version of any other content aggregator with all the associated problems. (The listicle and click-bait articles shown in the screenshot on the blog.mozilla link are very telling in that regard.) On the other hand, if the algorithms tailor content closely to a user's particular interests and views, that is the very definition of a filter bubble. [2]

Even putting aside these two worries, I find it deeply counterintuitive to expect automated content to better provide what a user wants to see than they are able to find themselves - e.g. with search engines. This is entirely counter to any previous experience I have had with automatic suggestions. I would very much like to see any justification for optimism here. And even if it were largely successful here, I don't see how it would avoid leaving very small sites, bloggers, and their ideas in the dust. I'm sceptical that a centralized source of recommendations is going to provide a solution to concentrated power on the web.

[1] This is the sort of line we got about recommendations from Pocket. "Bringing valuable content to the attention of users" is the kind of BS marketing-speak I expect from corps which run on ad revenue, but not from Mozilla.

[2] "A filter bubble is a state of intellectual isolation that can result from personalized searches when a website algorithm selectively guesses what information a user would like to see based on information about the user, such as location, past click-behavior and search history." - Wikipedia https://en.wikipedia.org/wiki/Filter_bubble

Not the job of a browser. But neither was DRM and an integrated search engine and yet here we are... There is a nice fork of FF for android that removes the Mozilla if anyone is interested

https://f-droid.org/en/packages/org.mozilla.fennec_fdroid/

Walls? When I jump over the 1 inch high wall and visit random sites, I'm risking getting spammed with modal popups and tracked and other shit. Yes I should use some script blocker, but they're too much effort to setup to avoid breaking things I need.

This is why I prefer to remain on a set of few trusted sites.

This seems to be one of those ideas everyone has had at some point, but that are hard to successfully implement.

I’d love to see it succeed. Don’t let the gratuitous negativity in this thread get to you.

this is the market working as intended.

search is overridden with advertising, corporate messaging, politics and other moneyed interests. in contrast, users are looking for authentic content and information. those two things are increasingly misaligned.

mozilla is experimenting with a new way for users to find what they want, rather than deferring to google et al. to deliver subpar user experiences.

it may work, it may not, but i’m glad they are trying.

(edit: i probably won’t try it any time soon, due to the privacy concerns voiced here and elsewhere. hopefully mozilla addresses those with a v.2 release and maybe i will try it then)

Nothing on the Laserlike website indicates how they plan to make money from their service, but I wouldn't bet on it not being overridden with advertising, corporate messaging, politics and other moneyed interests.

> this is the market working as intended

Is it? I would love for a transparent market-driven motivation to be presented. But in its absence, we can only guess.

> mozilla is experimenting with a new way for users to find what they want

Why? Because they don't like the existing ways? Because they believe it fits their mission? Because there is money on the line? My TV vendor also experiments with new ways to host apps and use the internet too.

> it may work, it may not, but i’m glad they are trying

I personally am not glad they are trying. In fact, these trials help solidify my beliefs in the Mozilla's ideals regarding the role of a browser and can potentially alienate me or at the least prevent them from maintaining an air of righteousness in my eyes.

you seem disillusioned by mozilla's actions. i can't help with that, other than to say that based on past actions, i believe mozilla will course-correct if necessary here.

i'm not defending this new "advance" experiment/feature (i'm reserving judgement until i know more about it). i'm only expressing appreciation for them trying to address a user pain point (that search is crappy), while also responding to the general question "why?" posited by the GP post.

i'll bite just a bit: you're concerned with sending your browser history to a third party but i assume are happy to send your search requests, hence search history, to a third party? i kinda understand the want to minimize the amount of third parties that know anything at all but otherwise don't see much of a difference in principal and history tells us that the search history may actually be worse.

The critical difference being that a search engine of any usefulness requires transmitting a search request to that third party. Pocket's recommendation feature does not require any such transmission, so it's definitely possible to work around.

yes, search history is juicy personal data. i'm willing to trust duckduckgo with that info for now. laserlike doesn't have a track record yet, so it's on the wait-and-see list.

if anyone's made an affordable personal search appliance yet, let me know. =)

> Second, the reasoning. Why?

Given how they've positioned Pocket and how Firefox's new tab page contains "Recommendations from Pocket", Mozilla seems to want you to enjoy browsing. They want you to enjoy using the Web and to enjoy it using Firefox. It's an interesting play and I'm excited to try it.

Strangely, all their ideas to heighten our enjoyment of browsing seem involve giving third parties privileged access to the browser.

Reply on news.ycombinator.com