Our results reveal no statistically significant reduction in ad-blocking or anti-tracking effectiveness for MV3 ad blockers compared to their MV2 counterparts, and in some cases, MV3 instances even exhibit slight improvements in blocking trackers.
3 Likes
That’s … kinda surprising. 
I think the biggest downside of the Ublock Origin’s MV3 version is its lack of dynamic filtering that would block significantly more unnecessary 3rd-party scripts compared to the default easy mode. If you’re not using it, the differences are smaller and probably something the user won’t notice.
3 Likes
According to UBO discusion thread on Github and This Reddit Thread,
uBO Lite:
Filter lists update only when the extension updates (no fetching up to date lists from servers)
-Many filters are dropped at conversion time due to MV3’s limited filter syntax
-No crafting your own filters (thus no element picker)
-No strict-blocked pages
-No per-site switches
-No dynamic filtering
-No importing external lists
-No support for entity concept, i.e. the replacement of effective TLD part with a wildcard. This causes rejection of many filters when converting to DNR rules.
-No support for redirect-if-blocked concept. This causes rejection of many filters making use of theredirect-ruleoption.
-No support for regex-based forredirect/transform/removeParams.
-No support for the concept of exception ofredirect/transformormodifyHeadersrules.
-No support for strict-partyness.
Also, in P.4 of the paper,
the MV3 update enforces a maximum of 30,000 blocking rules per extension while maintaining a collective restriction of 330,000 predefined blocking rules for all extensions in a browser, the latter raising concerns for ad blocker providers about the reduced effectiveness of a single MV3 ad blocker when used with multiple other MV3 ad blockers simultaneously.
While the paper suggest it does not, I wonder the list of site the used for testing, if they use a global list of sites, I suspect the conclusion will be very different.
3 Likes
You shouldn’t use two adblockers anyway, so I don’t see how this is an issue
Some think it is a possible counter measure for the “30k rules per extension” restriction by MV3.
Not me though, my counter measure to this 30k rules crap is to stay away MV3 enforced browsers.
1 Like
Its an open door for more advanced ad capability that is unblockable by the newest restricted tools.
For example, I’m pretty sure YouTube ads won’t be able to be blocked in V3 manifest.. Who know why …
Once there is a breach in the firewall, all ads will move to those methods.
I think the study was very solid. Especially the methodology, they put a lot of thought into each aspect of the process (adblocker selection, how to measure both anti-tracker effectivness and anti-ad, etc) and really explained it well, they even explained what a browser extension is.
And the results were very interesting. One notable finding is that in some cases, page with Stands (MV2) adblocker had more trackers and ads than without the extension, hinting that the extension inserted ads (my interpretation). This however, was gone with the MV3, because MV3 doesn’t allow such kind of practices (again, my interpretation).
It also showed that in term of tracking protection uBo>Adguard>Adblock + (ADBP)
Another interesting point: Only ADBP had roughly the same number of MV2 users than MV3 (40M and 45M), for the others MV3 was used by 100-1000 less users. Though Keep in mind that the study was performed in mid/late 2024.
They also performed the experiment on Firefox, which I really wanted to see because Firefox has superior MV2 capabilities than Chrome, so if MV2 can shine it is in Firefox. They only could analyse Ublock has it was the only MV3 and MV2 adblocker on Firefox (among the 4 selected). The results were similar on Firefox.
Now I do believe the study has one MAJOR flaw:
It only analyzed default filter lists and settings. While this is how most people will use it, tweaking can give you additional privacy.
At the very least, I would like to have seen a comparison between each adblocker with all* filterlists selected across MV2 and MV3.
*Excluding regional and perhaps also annoyance filters excluded.
Also, comparison with non-extreme blocking mode would have been useful, although more as a side-note cause you can’t really test those (as by definition they cause breakage hence in real-world people will have exceptions that will not be reflected in automatic testing).
1 Like
I want to counter this, they used a robust selection of widely used websites.
From the study:
For the main sample of our browser-based experiment, we selected ad-supported websites to observe the impact of ad blockers and evaluate their effectiveness. For this purpose, we utilized a curated list from HestiaLabs5, which comprised 195 websites. We selected these websites for their reliance on ad-supported content and their use of the Prebid framework, guaranteeing compatibility with the Azerion Ad Expert extension (see Section 5.2). This compatibility is crucial for ensuring the extension accurately records the number of ads, especially for the baseline browser instance using no ad blocker.
After manually reviewing each website to ensure the ad blockers and Azerion Ad Expert operated as expected, we refined the list to 114 websites. We excluded websites that redirected users, used ad-blocking prevention mechanisms, or were inaccessible. The final list for our main sample represents 114 popular, ad-supported websites across 10 countries.
We also conducted the browser-based experiment on two additional stratified samples for robustness tests. We used BuiltWith6 data of websites relying on the Prebid framework to conduct the stratified sampling: we stratified one sample by website employee count, yielding 191 websites, and another by Tranco7 popularity rank [48], yielding 185 websites. These stratified samples captured varying levels of organizational resources and popularity and are more representative of the overall population of ad-supported websites, enhancing the generalizability of our findings (see Section 6.4 and Appendix B.1).
1 Like
While most people connects to, for example, GAFAM, doesnt mean most people connects to them most of the time. And most of the world population dont live in Europe / US.
For myself, GAFAM contributes to less than 20% of my connections, and most of the GAFAM connects are actually youtube / GMS for notifications. Most my traffics actually goes to lots of different sites, which the company / people behind those sites are impossible to be included in their list.
The coverage of default blocklist is quite limited, thats why there are so many regional / specialised blocklists.
The research also selected the sites that compatible with the suites they use for analytics, and excluding redirect sites and sites that detects adblockers, I suspect it undermines the accuacy of the study. These are popular techniques to deter people from using adblockers / fool simple adblock rules.
Is the study consistent? Yes. Is it well defined? Yes. but does it reflect the real difference between Mv2 and MV3 blockers, I highly doubt it.
1 Like
This makes sense, those aren’t visited much by individuals
About adblock detection website:
If anything, it does bias toward “good behaviors” websites, but I fail to see how it impacts the difference between MV2 and MV3. The thing is that should they have included this, it could have impacted MV3 extensions and MV2 extensions differently, as some extension will not have been able to bypass this detection.
They haven’t only included GAFAM websites and have websites from 10 countries, which is a large sample.
Traffic in it self isn’t trackers though.
It’s about a framework of ad biding and not a specific vendor, but yes it is a limitation.
That was my main beef with the study, some lists might use advanced tracking protection or otherwise that aren’t in default lists.