Friday, July 24. 2009
What is this "Multisearch" thing in my Firefox about?
There is some valuable discussion about what the multisearch extension we just added to firefox in alpha3 does and why it does it the way it does it. Instead of replying to all the threads and bugs individually we decided that its better to put the information to a central place (like this blog).
So here the basic facts on what Multisearch is:
- First of all it’s an experiment designed for an ubuntu alpha release (meaning: this is not expected to make all happy, nor should someone expect this to be of production quality now).
- The multisearch extension adds a new search place to your browser experience, namely: newtab; hence the name “Multisearch”.
- The multisearch extension changes search identifiers in all four search places (homepage, awesomebar, quicksearch, newtab); also we had to switch to Google’s “Custom Search” as that was technically the most feasible way for us to get the anonymous usage patterns we are looking for in this experiment.
- The multisearch extension can be disabled in Tools -> Addons; this was done intentionally so you can easily (and should) do this if the Custom Search interferes with your workflow.
So that’s basically what it does; imo, there is not much else to wonder ‘bout, really! Also there is by far no conspiracy, nor do we want to remove features from your google search; it’s just that for now we want the data for your usage
patterns mentioned in 3.
Anyway, I don’t want to reply to each and every claim and rumor and conspiracy theory posted around the web; but still I want to remove some understandable concerns that sprung up. So here a few more points that I think are worthwhile to mention:
- “Multisearch” (extension id: me001@canonical.com) is not intended to stay forever – at least not in its current form.
- “Multsearch” 0.0.1 does not reflect how a newtab experience would look like, if we decide to implement something like that for karmic; check out the mozilla newtab efforts to get an idea [1][2]; something like that combined with our own ideas would probably be closer to any eventual karmic newtab experience. If you like the general idea of having a useful newtab experience and want to contribute, use this event as a chance to file bugs about it. Remember to tag your bug “multisearch”, as otherwise we might not see it when reviewing all the feedback we got.
- “Multisearch” is not a well thought out name ;); it was just choosen as a project codename; we didn’t want to put much thought in it for now; also I felt using “Multisearch” is a better name than just “me001@canonical.com”.
Last but not least, I want to answer a few core concerns I picked up from the various places I found:
- “You installed this without asking/permission”: We regularly change features for software during the development release; also we add new stuff to our default installs that will get automatically installed if you opted into ubuntu-desktop; I agree that it might have been better to move this to a standalone package and seeding that through ubuntu-desktop; but then its just an intermediate thing what you see now and you can always disable it in Tools -> Addons for the time being.
- “You are trying to make money from me”: For now nobody wants to make money from you. If we wanted to make money putting this in an early alpha release would probably be a pretty bad business decision. Even if we wanted to make money from you, we should still clarify what “we” means here; it basically means “ubuntu”: and even though it might practically mean canonical its obvious that getting a constant revenue stream will help to fund more talented developers and events to keep ubuntu the best distribution that exists.
- “Custom Search delivers an inferior search experience”: this is true, but using the “Custom Search” bits was the most feasible way of pursuing this experiment for us; we are also sure that this experiment will be of a great longterm benefit for the ubuntu project as a whole, so please be patient.
[2] – http://hg.mozilla.org/users/dmills_mozilla.com/about-tab/
Hi Alex.
I already told you this, but would like to reiterate it:
I would not have such a prob with this if you just added a new search and made it default, allowing the user to reset to Google search easily.
The most important questions have not been answered IMnsHO. Yes, features are added and/or changed all the time but that is not what the commotion is about. This "Multisearch" thing a) is used to collect data about web searches without permission or notification and b) throws a bunch of ads in the prime-time visual space of the search results page, also unrequested. The former is a blatant violation of several laws in many countries, the latter is just bad form. Why did you depart from the previous sensible approach to gathering user data as implemented by eg. Popularity Contest? Ask the user if they agree to their search data being collected and give them the opportunity to help the project by showing them some ads. Don’t hide these activities. Things which are hidden are usually not meant to be seen. You are playing with your user’s trust in the project. It is very easy to lose it, very hard to regain.
Some things I’ve been curious about that have not been addressed in all of the posts/bugreports…
Is this multisearch related to or derived from either of the multisearch extensions listed on the mozilla add-ons page or is this a totally new project? The version number matches a several year old version of one of them but I don’t want to jump to any conclusions as 0.0.1 is a fairly generic version number.
I appreciate that the extension can be disabled, but it is still not removable from the Add-Ons screen. Arguably, I could remove the files manually, but I really shouldn’t have to do that to be rid of it.
I can accept the implementation in the main package as a way to get a wider testing pool in the alpha, that’s not really a huge deal. The sticking point for me is it took nearly a week to get really ANY information on what the multisearch is, and what it does. The information should have come well before the release, especially with something that you admit is for statistics gathering.
Which brings me to the thing that’s glaringly absent… What are you planning on doing with the statistics? What is the logical end of this endeavour?
It’s clear that many users are deeply concerned about the kind of ‘anonymous usage data’ that is being collected by Multisearch. Your public responses, Alex, have in no way addressed what data is collected, how it is collected, and what it is used for. No notification was given that data – anonymous or otherwise – would be collected (even Microsoft have an opt-in for anonymous data collection in IE!)
The argument that the extension can be disabled via the Addons dialogue is disengenuous – a user who is not informed of the purpose or action of the extension has no understanding of whether or not it should be disabled.
The fact that Multisearch changed "search identifiers in all four search places" significantly affects users outside the US who may rely on localised searching, and to do this without any notification at all, even in an Alpha release, speaks of a contempt for the hundreds – if not thousands – of volunteer testers who work on making Ubuntu releases functional for other end-users. And that, Alex, is rude.
> Your public responses, Alex, have in no way addressed what data is collected, how it is collected, and what it is used for
We only look at the relative usage ratio of the four search places mentioned; and I strongly disagree that there is any privacy impact. We only get aggregated results and we only look at data at an even higher aggregated level.
Multisearch may be experimental, but the custom search has been there since hardy.
So please let’s stop mixing the two things. The custom search is not an experimental feature. And it looks like unauthorised spyware.
Even assuming its benign, the user experience is horrid. Plain horrid.
Personally, I just plain don’t like the mass of adverts a user has to wade through to get to the search results.
I’d be happy if the adverts were unobtrusive, but they’re ugly as hell. Maybe people would be more forgiving if you’d actually add some value to the Google search experience at the same time as you’re using it as a revenue generation tool. Maybe people would be more forgiving if you had been honest from the start, and got user buy-in, rather than trying to foist this on people.
While this is a test stage, you should look long and hard at what you’re doing. If there is not an opt-in rather than an opt-out option, you’re going to lose the people who use Firefox due to their philosophies. That’s your core user base, the rest get their information and recommendations from those core users.
I don’t like the fact you don’t see the normal google search results with web/image/video at the top – I’m sorry, I’d prefer just to have a normal google search from the search box at top/right. Feel free to add another option to the search type in that box, but I’d like the normal google one to do a normal google search
Is there a way that Ubuntu can do something like what Mozilla does, where Mozilla gets adwords revenue but gives you "normal" google?
Hi,
Changing the default behaviour of the firefox AwesomeBar and the search box is stupid, wrong, and obnoxious.
The main reason that I use firefox, and I know this is also a major reason that many of my users choose to use firefox, is that it behaves the same way on all of my computers.
It is absolutely the most important feature of firefox for linux that it behaves exactly the same way as firefox for windows and firefox for MacOS. I don’t give a damn about how your changes make it look like canonical is taking over people’s browsing experience, or whatever. I care that when I type something in the awesomebar on my mac, it goes to http://www.google.com/search?ie=UTF-8&oe=UTF-8&sourceid=navclient&gfns=1&q=[query], and if I then want to switch to image results, I can click on "images".
If this thing stays in ubuntu to karmic stable, I will be reimaging people’s linux desktops with fedora. While it would probably take more time than upgrading to karmic, it will be worth the effort of not having to log in as every user who might use a machine to disable a firefox addon.
Just my two cents, if you will.
It’s on slashdot now … think about all the negativity this will attract. You just brought this onto yourselves.
so you want to make money off my clicks? then you had better come over with your lawyers so we can draw up a contract and you guys can make arrangements to give me 50%. otherwise you are stealing something from me and selling it for a profit.
If Canonical finds this kind of douchebaggery to be acceptable, especially to those who are contributing to the alpha testing, could they publicly state as such?
I am seriously fucking incredulous that anyone – and I’m looking DIRECTLY AT YOU ALEXANDER - would consider this acceptable to the open source community, the key word there being "open".
If you want to leech our usage data for your own nefarious purposes, at least grow some fucking balls and forewarn us. The "oh but you can disable it!" argument is absolute bullshit and you know it.
This is Crap. You know it too. Stop equivocating. Stop trying to justify bad behavior because your a FOSS project.
When Microsoft did this, it was crap: http://voices.washingtonpost.com/securityfix/2009/05/microsoft_update_quietly_insta.html
When you do it, should you get a free skate because it’s a FOSS project? No. You should not. Doing anything like this will always be viewed as underhanded, slimy and lumped in with the behavior of cancerous corporate entities. Rightfully so. You have no right to gather anything, or install anything on a private system until you meet all the checks and balances. Namely:
1) Inform the user what you are doing.
2) Give them a chance to opt out before you flip one switch on a disk platter.
You must meet those checks and balances, or you run the risk of violating the trust of the people you wish to win over. You met none of those. That’s why you get the Microsoft comparisons. That’s why people are pissed off. They have a right to be. When you act like a slimy underhanded cretin, and then attempt to justify it because you refuse to leave your own perspective to step into the shoes of the users you claim to serve…you deserve all the negative feedback you get.
You should remove this. Then, you and Mark Shuttleworth should follow up with a mea culpa, an apology of extreme proportions, possibly hire skywriters so the message gets out that you are sorry for what you did, you will never do it again, and please, would everyone forgive you.
I recently was considering different distros for my main OS, and though I did in the end go with Debian testing, I was pretty close to going with Ubuntu as a choice. (I liked it as DomU guest os in the Xen live disc.)
But I do have to say Alex, that this idea, that of getting those who are testing your software, or those who like to live on the bleeding edge of development at all times, to test this without informing them it is there, is not a positive development.
The whole FOSS philosophy is built on trust and reasonableness, and such a change to the default behaviour of a major tool in any OS is not a positive one. It’s made far worse by the fact that you did not bother to announce this to those who intended on testing such builds.
Saying, well its no problem as it’s just a test and we do this all the time, changing features etc, is simultaneously true, as in yes you have that right, but also horrible in this context.
I think if you had put out a plugin and told people,’ hey guys, we get Revenue if you use this particular search plugin, it’ll help keep us strong’ most of your userbase would probably have used it on general principle. Right now though, you have risked all of that strength of feeling by demonstrating a willingness, in however a narrow context as you may wish to see it, to make the implementation non-consensual. and guess what, the community at large noticed it and published it.
Ubuntu has gathered a huge amount of trust in recent years from its user base, and the community at large for producing a very nice GNU/Linux desktop, but with un-discussed actions of this nature you risk it all, people are rightly very sensitive to Spyware and companies that seem to think that Spyware is ‘okay’ because they have no bad intentions (For now..).
Personally, I think you ought to give serious consideration to how this whole ‘linux’ shebang came about. It’s because of trust and a willingness to help voluntarily.
RMS, Linus and co. didn’t put a gun to anybody’s head to get to where we are today, they simply asked that people of a similar mindset help them. Please go back to that philosophy and move on from there.
Wow. Your alpha and beta testers deserve an apology. Why is it so hard to see that?
How is this not spyware? Its an application that tracks useage patterns of users. How much more clear cut does it get?
Frankly, this could have been more transparent; the glut of mis-informed people here are definitely the fallout from that.
If you install an alpha release of an Operating System, there are things in it that aren’t going to work right. Period.
If this was announced on a developer’s list or "hey this is something we’re testing in Karmic" and practically made required reading by anyone installing the Karmic alpha, this would have been a non-issue instead of the ridiculous flaming and wild conspiracy theorizing that’s taken place.
In the FLOSS world you can’t beg for forgiveness: you have to ask for permission.
Please be better about this in the future.
(I didn’t install Karmic alpha3/4.)
If it had been announced as "hey this is something were testing in Karmic" maybe with instruction for removal, nobody would have complained, but ubuntu would have gathered much less data. The impression here is that the extension was NOT advertised to end-users on purpose, that’s the problem.
honestly, personally, I don’t believe in the "if you had announced this up-front nobody would have reacted like this" argument. Point is that most complainers wouldn’t have read that announcement – similar to how they now don’t spot the important points in the available content we brought up in a timely fashion. Also without any hard facts there usually is even more room for speculation, which heats the discussion even further.
Agreed. Had this been presented as it should have been, many aspects of this would have been immediately shot down by the community based on the same principles that are fueling this fire now.
Like stated, Canonical owes the testers an apology
I don’t think we need to apologize. However, we are – as always – really thankful for all the alpha testers we have.
Also, I talked to many of them in person about this experiment, and most didn’t see any issue at all. The others said we could have been more diligent about communicating this up front. Looking back I might agree with that, though i don’t really believe that it would have made a big difference.
Anyway, thanks for caring.
I, personally, don’t have a problem with this. It’s alpha software that would have probably been removed anyway. Canonical has always been good about putting everything that changes in their notes.
No complaints here.
What ever happened to the users – (Where’s TRON when we need him?) Anything that sends information OUT of my computer that I have not pre-authorized is in violation of many privacy policies, regardless of the aggregate. I am aware of much that goes out (IP address for one) but to include something as this that is UNNECESSARY is both inappropriate and an affront to the intelligence of the FOSS community.
Let this be a wake-up call not to "assume" that everyone in the FOSS community is benevolent and good.
So have you now discovered the bleeding obvious from your little statistical exercise: users like about:blank on new tabs, often set the home page to the same thing, and hate focus-stealing search boxes on home pages?
Oh, there is at least one good feature in this multisearch plugin: You can disable it!
It really totally ruins the otherwise awesome Firefox experience I have so much used to. Best thing in Firefox is its absolutely perfect address bar which integrates nicely with Google (calculator/converter, "feeling lucky", auto-completion,image search etc. and of cource all in Finnish too).
Otherwise, I have been really pleased with Karmic Koala. This thing works perfectly with Intel graphics thanks to KMS. Single most annoying thing among previous Linux distributions was slooow resume from suspend. Now it’s quite instant.
So please, remove this "feature" completely, and if you want to gather user experiences or do usability testing, please inform us and ask permissions. Thank you!
/me on Identi.ca
asac: everyone welcome @micahg to identi.ca - a rocking !ubuntu / #mozillateam contributor
asac: arrived @ home ... getting in from international through frankfurt is painful - lufthansa could surely do better with some common sense
asac: forgot to charge my mini 9 for travel ... lets hope i can find time to do that a bit in seattle before going on long distant flight
asac: PDX airport + free wifi: nice!
asac: one more day to go in portland - #sprint
asac: done: discussed desktop-webmail with @mvo
asac: bad packaging practice is clearly to NOT use embedded tarballs for huge upstream trees - #fail #nobodygetsitright
