Hey Josh,

Thanks for responding.   Why was the OT stopped before stable?  How many 
people are in each of the channels (canary/dev/beta/stable)?

Cheers,

Geoff

On Friday, September 10, 2021 at 1:07:42 PM UTC-6 Josh Karlin wrote:

> Up to 0.5% max, and it was enabled for canary/dev/beta channels, but not 
> stable.
>
> On Fri, Sep 10, 2021 at 2:59 PM Geoff Yaworski <[email protected]> 
> wrote:
>
>> For historical sake.  Where did the ramp up max out at?  I don't see an 
>> update on this, so if there is an official place to look, please let me 
>> know.  By where this thread left out, sounds like it was at 0.5%.
>>
>> On Friday, April 2, 2021 at 11:40:49 AM UTC-6 Josh Karlin wrote:
>>
>>> Note that the rollout has started. Currently, the interestCohort API 
>>> will work for 50% of canary, dev, and beta channel users if you have an 
>>> Origin Trial token. The plan is to ramp up stable to 0.5% if we don't come 
>>> across any issues early next week. From there we're exploring options for 
>>> ramping up to 5%. 
>>>
>>> Again, once you have an Origin Trial token, only a subset of users will 
>>> actually have the interestCohort() API enabled, so be sure to use feature 
>>> detection. The reason that we limit the population is so that you don't 
>>> have to worry about using the API on more pages than is permitted for 
>>> Origin Trials (0.5% of page loads by default). 
>>>
>>> Josh
>>>
>>> On Thu, Apr 1, 2021 at 8:01 PM Federico Nieves <[email protected]> 
>>> wrote:
>>>
>>>> Hey Yao/Michael,
>>>>
>>>> Thanks it worked, the thing is that if you launch Chrome from the CLI 
>>>> with the parameters, you have to make sure Chrome is completely closed 
>>>> before. Otherwise the flags won't take effect.
>>>>
>>>> For the community, this is a good URL to test if the browser has Flocs 
>>>> enabled: https://floc.glitch.me/
>>>>
>>>> Best,
>>>> On Thursday, April 1, 2021 at 7:50:09 PM UTC-3 Yao Xiao wrote:
>>>>
>>>>> Hi Federico,
>>>>>
>>>>> I believe you want to test in a https page. The API additionally 
>>>>> requires to be called in a "secure context", and http://localhost 
>>>>> isn't considered to be one at this point.
>>>>>
>>>>>
>>>>> On Thursday, April 1, 2021 at 6:41:37 PM UTC-4 Michael Kleber wrote:
>>>>>
>>>>>> The current version of https://web.dev/floc/ lists all the flags you 
>>>>>> would need to test on your own browser.
>>>>>> https://web.dev/floc/#as-a-web-developer-how-can-i-try-out-floc
>>>>>> You'll need:
>>>>>>
>>>>>> --enable-blink-features=InterestCohortAPI 
>>>>>>
>>>>>> --enable-features="FederatedLearningOfCohorts:update_interval/10s/minimum_history_domain_size_required/1,FlocIdSortingLshBasedComputation,InterestCohortFeaturePolicy"
>>>>>>
>>>>>> Once you're using those flags, you don't need an origin trial token; 
>>>>>> the flags take care of it all.
>>>>>>
>>>>>> --Michael
>>>>>>
>>>>>>
>>>>>> On Thu, Apr 1, 2021 at 6:21 PM Federico Nieves <
>>>>>> [email protected]> wrote:
>>>>>>
>>>>>>> Hi everyone,
>>>>>>> I would like to start testing flocs on my localhost, however I 
>>>>>>> always receive the error message: "document.interestCohort is not a 
>>>>>>> function". 
>>>>>>>
>>>>>>> On about:flags there is no option to enable flocs, however as per 
>>>>>>> Abhinav comment, I launched it with:
>>>>>>> > MacOS : open -na Google\ Chrome --args 
>>>>>>> --enable-blink-features=InterestCohortAPI
>>>>>>> I also registered a new token for localhost.
>>>>>>>
>>>>>>> Is there anything I'm missing here?
>>>>>>>
>>>>>>> Thanks in advance.
>>>>>>>
>>>>>>>
>>>>>>> On Friday, March 19, 2021 at 8:25:37 PM UTC-3 [email protected] 
>>>>>>> wrote:
>>>>>>>
>>>>>>>> I was left unsure about how we can see any impact. Can we enroll 
>>>>>>>> participants to be in the group? How can we measure the impact?
>>>>>>>>
>>>>>>>> On Friday, March 19, 2021 at 8:05:19 AM UTC-7 Greg Truchetet wrote:
>>>>>>>>
>>>>>>>>> Thank you Josh, seems good :)
>>>>>>>>>
>>>>>>>>> Another question, do you have a dedicated group for talking about 
>>>>>>>>> the running trial or can we use this thread?
>>>>>>>>> The feature activation with the token seems to work fine now but 
>>>>>>>>> once activated, we get this error: document.interestCohort is not a 
>>>>>>>>> function
>>>>>>>>>
>>>>>>>>> Here is the concerned piece of code running on Chrome 89 instances:
>>>>>>>>> const isChrome = !!window.chrome && (!!window.chrome.webstore || 
>>>>>>>>> !!window.chrome.runtime);
>>>>>>>>> if (isChrome && !!document.featurePolicy && 
>>>>>>>>> !!document.featurePolicy.features() 
>>>>>>>>> && document.featurePolicy.features().includes('interest-cohort')) 
>>>>>>>>> {
>>>>>>>>>   return await document.interestCohort() || {};
>>>>>>>>> }
>>>>>>>>>
>>>>>>>>> Are we missing something? :)
>>>>>>>>>
>>>>>>>>> Thanks again for your help!
>>>>>>>>>
>>>>>>>>> On Friday, March 19, 2021 at 1:45:27 PM UTC+1 Josh Karlin wrote:
>>>>>>>>>
>>>>>>>>>> Hi Greg. Thanks for pointing that out. Should be fixed now.
>>>>>>>>>>
>>>>>>>>>> Josh
>>>>>>>>>>
>>>>>>>>>> On Thu, Mar 18, 2021, 1:24 PM Greg Truchetet <[email protected]> 
>>>>>>>>>> wrote:
>>>>>>>>>>
>>>>>>>>>>> Hi guys,
>>>>>>>>>>>
>>>>>>>>>>> (not sure my previous message worked,  new try ;) )
>>>>>>>>>>> We have noticed that FloC is now available on the Chrome trials 
>>>>>>>>>>> page, but the third-party option is not present, unlike 
>>>>>>>>>>> conversion-measurement.
>>>>>>>>>>>
>>>>>>>>>>> Is it intended or is it because the rollout of the trial is not 
>>>>>>>>>>> finished yet? :)
>>>>>>>>>>>
>>>>>>>>>>> Thanks for your help
>>>>>>>>>>>
>>>>>>>>>>> On Monday, March 15, 2021 at 10:38:15 PM UTC+1 
>>>>>>>>>>> [email protected] wrote:
>>>>>>>>>>>
>>>>>>>>>>>> Hi API owners, 
>>>>>>>>>>>>
>>>>>>>>>>>> We'd like to request an increase from the standard “0.5% of 
>>>>>>>>>>>> page loads” origin trial usage target to 5%, because 0.5% of data 
>>>>>>>>>>>> would be 
>>>>>>>>>>>> too small for ad tech to train a sound ML model / to evaluate the 
>>>>>>>>>>>> effectiveness of the FLoC. 
>>>>>>>>>>>>
>>>>>>>>>>>> For whatever threshold that's finally allowed, we'll configure 
>>>>>>>>>>>> a user sampling to ensure it doesn't cross the OT threshold. 
>>>>>>>>>>>> On Wednesday, March 10, 2021 at 2:29:11 AM UTC-5 
>>>>>>>>>>>> [email protected] wrote:
>>>>>>>>>>>>
>>>>>>>>>>>>> Thanks for the details.
>>>>>>>>>>>>>
>>>>>>>>>>>>> Regards,
>>>>>>>>>>>>> Abhinav
>>>>>>>>>>>>>
>>>>>>>>>>>>> ------------------------------
>>>>>>>>>>>>> *From:* Yao Xiao <[email protected]>
>>>>>>>>>>>>> *Sent:* Wednesday, 10 March, 2021, 2:38 am
>>>>>>>>>>>>> *To:* blink-dev
>>>>>>>>>>>>> *Cc:* Abhinav Sinha; Yao Xiao; Josh Karlin; Michael 
>>>>>>>>>>>>> "Chromium" Kleber; Marshall Vale
>>>>>>>>>>>>> *Subject:* Re: Intent to Experiment: Federated Learning of 
>>>>>>>>>>>>> Cohorts
>>>>>>>>>>>>>
>>>>>>>>>>>>>
>>>>>>>>>>>>> Hi,
>>>>>>>>>>>>>
>>>>>>>>>>>>> The document.interestCohort API will be made available in a 
>>>>>>>>>>>>> third-party Origin Trial (
>>>>>>>>>>>>> https://web.dev/third-party-origin-trials/). The Origin Trial 
>>>>>>>>>>>>> hasn't started yet and we expect it to start in the middle of 
>>>>>>>>>>>>> M89. When it 
>>>>>>>>>>>>> starts, the API will be available on sites that sign up for the 
>>>>>>>>>>>>> OT, subject 
>>>>>>>>>>>>> to region restrictions and 5% user sampling.
>>>>>>>>>>>>>
>>>>>>>>>>>>> For your questions specifically,
>>>>>>>>>>>>> 1. It’s available equally on all sites, for first or third 
>>>>>>>>>>>>> parties that participate in the Origin Trial.
>>>>>>>>>>>>> 2. Ad tech are supposed to observe the browsing behavior of 
>>>>>>>>>>>>> floc ids, and train their models based on it.
>>>>>>>>>>>>> 3. This API will initially be available for 5% of users in the 
>>>>>>>>>>>>> selected regions. The user sampling is to ensure that the API 
>>>>>>>>>>>>> usage doesn’t 
>>>>>>>>>>>>> reach beyond OT thresholds. 
>>>>>>>>>>>>>
>>>>>>>>>>>>> On Tuesday, March 9, 2021 at 10:44:37 AM UTC-5 
>>>>>>>>>>>>> [email protected] wrote:
>>>>>>>>>>>>>
>>>>>>>>>>>>>> As per the qualifying user's definition 
>>>>>>>>>>>>>> <https://github.com/WICG/floc#qualifying-users-for-whom-a-cohort-will-be-logged-with-their-sync-data>
>>>>>>>>>>>>>>  I 
>>>>>>>>>>>>>> do not see the feature & JS API (document.interestCohort) 
>>>>>>>>>>>>>> working in the 
>>>>>>>>>>>>>> latest chrome release 89.0.4389.82 ( Tested US and APAC region)
>>>>>>>>>>>>>> The chrome feature status 
>>>>>>>>>>>>>> <https://www.chromestatus.com/feature/5710139774468096> also 
>>>>>>>>>>>>>> states this as behind developer flag. I was able to force enable 
>>>>>>>>>>>>>> API using 
>>>>>>>>>>>>>> blink feature but doesn't seem to be ideal for OT.
>>>>>>>>>>>>>>
>>>>>>>>>>>>>> Can we please elaborate on the Origin Trial scope and 
>>>>>>>>>>>>>> coverage around following:
>>>>>>>>>>>>>>
>>>>>>>>>>>>>>    1. Is this FloC data available by default to only 
>>>>>>>>>>>>>>    Chrome/Google at scale through account sync?
>>>>>>>>>>>>>>    2. How can any publisher , ssp , advertiser analyse the 
>>>>>>>>>>>>>>    floc id data at scale during OT phase?
>>>>>>>>>>>>>>    3. Will this ID / API be available for all qualifying 
>>>>>>>>>>>>>>    users for selected regions or there are additional sampling / 
>>>>>>>>>>>>>> selection 
>>>>>>>>>>>>>>    criteria ?
>>>>>>>>>>>>>>
>>>>>>>>>>>>>> Link to same query on GitHub - 
>>>>>>>>>>>>>> https://github.com/WICG/floc/issues/57 
>>>>>>>>>>>>>>
>>>>>>>>>>>>>> On Thursday, March 4, 2021 at 3:00:19 AM UTC+5:30 
>>>>>>>>>>>>>> [email protected] wrote:
>>>>>>>>>>>>>>
>>>>>>>>>>>>>>> *Contact emails*
>>>>>>>>>>>>>>> [email protected], [email protected], [email protected], 
>>>>>>>>>>>>>>> [email protected]
>>>>>>>>>>>>>>
>>>>>>>>>>>>>>
>>>>>>>>>>>>>>>
>>>>>>>>>>>>>>> *Explainer*
>>>>>>>>>>>>>>> https://github.com/WICG/floc
>>>>>>>>>>>>>>>
>>>>>>>>>>>>>>> *Summary*
>>>>>>>>>>>>>>> The FLoC API intends to provide callers (primarily ad-tech) 
>>>>>>>>>>>>>>> with a "cohort label" shared by thousands of people with 
>>>>>>>>>>>>>>> similar browsing 
>>>>>>>>>>>>>>> habits to aid in interest-based online advertising. The browser 
>>>>>>>>>>>>>>> will group 
>>>>>>>>>>>>>>> people together into k-anonymous interest cohorts via purely 
>>>>>>>>>>>>>>> on-device 
>>>>>>>>>>>>>>> computation, and reveal the user’s cohort via script.
>>>>>>>>>>>>>>>
>>>>>>>>>>>>>>> We would like to launch via an Origin Trial to start getting 
>>>>>>>>>>>>>>> developer feedback.
>>>>>>>>>>>>>>>
>>>>>>>>>>>>>>>
>>>>>>>>>>>>>>> *Link to “Intent to Prototype” blink-dev discussion *
>>>>>>>>>>>>>>> https://groups.google.com/a/chromium.org/g/blink-dev/c/DpZZG5K1PWs/m/IJT1W69wCAAJ
>>>>>>>>>>>>>>>
>>>>>>>>>>>>>>> *Goals for experimentation*
>>>>>>>>>>>>>>> We hope to get feedback from ad tech suggesting whether FLoC 
>>>>>>>>>>>>>>> is viable.
>>>>>>>>>>>>>>>
>>>>>>>>>>>>>>> We’ll keep monitoring the anonymity and sensitivity 
>>>>>>>>>>>>>>> properties based on UKM data, and will trigger the blockings of 
>>>>>>>>>>>>>>> some 
>>>>>>>>>>>>>>> cohorts or a full assignment / recomputation as necessary.
>>>>>>>>>>>>>>>
>>>>>>>>>>>>>>> *Experimental timeline*
>>>>>>>>>>>>>>> We expect to start the Origin Trial in M89 (Stable on Mar 2, 
>>>>>>>>>>>>>>> 2021)
>>>>>>>>>>>>>>>
>>>>>>>>>>>>>>> *Risks*
>>>>>>>>>>>>>>>
>>>>>>>>>>>>>>>    - *Interoperability and Compatibility* 
>>>>>>>>>>>>>>>       - No signals from other browsers.
>>>>>>>>>>>>>>>    - *Activation*
>>>>>>>>>>>>>>>    - The API is straightforward to use but understanding 
>>>>>>>>>>>>>>>       how to interpret the meaning of an interest cohort may be 
>>>>>>>>>>>>>>> complex.
>>>>>>>>>>>>>>>       - *Privacy* 
>>>>>>>>>>>>>>>       - The current implementation of FLoC for this Origin 
>>>>>>>>>>>>>>>       Trial creates a new fingerprinting surface in that it 
>>>>>>>>>>>>>>> provides the same 
>>>>>>>>>>>>>>>       value across sites. However, it provides a far smaller 
>>>>>>>>>>>>>>> fingerprinting 
>>>>>>>>>>>>>>>       surface than the APIs it’s intending to replace (third 
>>>>>>>>>>>>>>> party storage). 
>>>>>>>>>>>>>>>       We’re looking at ways to reduce it further down the line 
>>>>>>>>>>>>>>> (e.g., privacy 
>>>>>>>>>>>>>>>       budget).
>>>>>>>>>>>>>>>       - The methods for ensuring that each cohort is large 
>>>>>>>>>>>>>>>       enough and not correlated with sensitive categories are 
>>>>>>>>>>>>>>> described in the 
>>>>>>>>>>>>>>>       explainer 
>>>>>>>>>>>>>>>       <https://github.com/WICG/floc#proof-of-concept-experiment>
>>>>>>>>>>>>>>>       .
>>>>>>>>>>>>>>>       - The long-term intent is for FLoC calculation to be 
>>>>>>>>>>>>>>>       fully opt-in at the page level (e.g., pages that use the 
>>>>>>>>>>>>>>> API would be 
>>>>>>>>>>>>>>>       included in FLoC calculation). For an Origin Trial where 
>>>>>>>>>>>>>>> only a small 
>>>>>>>>>>>>>>>       sample of page loads will use the API, it will be 
>>>>>>>>>>>>>>> necessary to include more 
>>>>>>>>>>>>>>>       sites in the cohort calculation in order to be useful. 
>>>>>>>>>>>>>>> So, for the OT only, 
>>>>>>>>>>>>>>>       we will include pages in cohort calculation that we 
>>>>>>>>>>>>>>> anticipate will use 
>>>>>>>>>>>>>>>       FLoC in the future, e.g., pages that have ad resources as 
>>>>>>>>>>>>>>> determined by 
>>>>>>>>>>>>>>>       Chrome’s Ad Tagging 
>>>>>>>>>>>>>>>       
>>>>>>>>>>>>>>> <https://chromium.googlesource.com/chromium/src/+/master/docs/ad_tagging.md>
>>>>>>>>>>>>>>>       .
>>>>>>>>>>>>>>>       - We also provide an interest-cohort permissions 
>>>>>>>>>>>>>>>       policy for pages to opt out from FLoC completely.
>>>>>>>>>>>>>>>       - End users who have disabled third-party cookies in 
>>>>>>>>>>>>>>>       or before M89 will not be included in the experiment. In 
>>>>>>>>>>>>>>> M90 we intend to 
>>>>>>>>>>>>>>>       introduce a separate end user control.
>>>>>>>>>>>>>>>       - This Origin Trial will be available in selected 
>>>>>>>>>>>>>>>       regions.
>>>>>>>>>>>>>>>       - Please see the 
>>>>>>>>>>>>>>>       
>>>>>>>>>>>>>>> https://github.com/WICG/floc#privacy-and-security-considerations
>>>>>>>>>>>>>>>  
>>>>>>>>>>>>>>>       and 
>>>>>>>>>>>>>>>       
>>>>>>>>>>>>>>> https://github.com/WICG/floc/blob/master/security-and-privacy-self-review.md
>>>>>>>>>>>>>>>  
>>>>>>>>>>>>>>>       for more privacy questions & answers.
>>>>>>>>>>>>>>>    
>>>>>>>>>>>>>>> *Any risks when the experiment finishes?*
>>>>>>>>>>>>>>> No. When the experiment finishes, the 
>>>>>>>>>>>>>>> document.interestCohort API will stop being exposed, but the 
>>>>>>>>>>>>>>> site is 
>>>>>>>>>>>>>>> expected to handle that due to the OriginTrial only being 
>>>>>>>>>>>>>>> enabled for some 
>>>>>>>>>>>>>>> page visits.
>>>>>>>>>>>>>>>
>>>>>>>>>>>>>>> *Reason this experiment is being extended*
>>>>>>>>>>>>>>> N/A
>>>>>>>>>>>>>>>
>>>>>>>>>>>>>>> *Ongoing technical constraints*
>>>>>>>>>>>>>>> No
>>>>>>>>>>>>>>>
>>>>>>>>>>>>>>> *Will this feature be supported on all five Blink platforms 
>>>>>>>>>>>>>>> supported by Origin Trials (Windows, Mac, Linux, Chrome OS, and 
>>>>>>>>>>>>>>> Android)?*
>>>>>>>>>>>>>>> Yes
>>>>>>>>>>>>>>>
>>>>>>>>>>>>>>> *Link to entry on the feature dashboard 
>>>>>>>>>>>>>>> <https://www.chromestatus.com/>*
>>>>>>>>>>>>>>> https://www.chromestatus.com/features/5710139774468096 
>>>>>>>>>>>>>>>
>>>>>>>>>>>>>>>
>>>>>>>>>>>>>
>>>>>>
>>>>>> -- 
>>>>>> Forewarned is worth an octopus in the bush.
>>>>>>
>>>>>>

-- 
You received this message because you are subscribed to the Google Groups 
"blink-dev" group.
To unsubscribe from this group and stop receiving emails from it, send an email 
to [email protected].
To view this discussion on the web visit 
https://groups.google.com/a/chromium.org/d/msgid/blink-dev/ad55234c-f08a-4fd5-96a4-a8284db34964n%40chromium.org.

Reply via email to