Thread by @DanielHolznagel

#DSA - Et Voila – finally a 600pages document for the final draft oft he #DSA (Trilogue settled version + technical amnds) has found its way to the public.

@TspBackgroundDi and there tech-expert @Torbatschow reports about it.

Now finally we can look at details!

long thread 1/X

Overall impression: Council position prevailed in many instances over EP (good!)

After looking at the document, here my 50cent on points that I found still unclear after the Trilogue so far (will leave out points clear from press releases – this here for afficionados):

2/X

Recital 13: cloud and technical web hosting out of the DSA!? „cloud computing ... and web hosting ... when serving as infrastructure, such as the underlining infrastructural ... services ... should not in themselves be considered as disseminating to the public …“

3/X

Recital 14: Telegram in? „… private messaging services, fall outside the scope …However, the obligations …may apply to services that allow the making available of information to a potentially unlimited number of recipients,... as through public groups or open channels.“

4/X

Recital 18: unfortunatly, no progress on the definition of „active role“, which will fail to codify ECJ YouTube/Cyando, as described here, page 5:
hateaid.org/wp-content/upl…

At least, Recital 20 of Council-position will come (deliberate collaboration leads to active role).

5/X

Recital 33 (most underrated) untouched. That is: orders by authorities/courts „do not in principle restrict those providers’ freedom to provide their services across borders“. This is a real new hole in the country-of-origin-principle.

6/X

Art. 2(g): definition of „illegal content“ basically unchanged. Problem: illegal is content, when „not in compliance with … the law of a Member State“ … so strictest national laws will rule for the whole Union? Holocaust denial illegal everywhere under the DSA?

7/X

No surprises with Art. 6 (good samaritan) … the Article will have no pracitcal effect as explained elsewhere (CR 2021, 123)

Art. 7 (no-monitoring): Thanks god, EP amendments rejected, stays Commission text basically...

8/X

Future will see courts here active in interpreting, see e.g. Künast v. Facebook requiring Upload-Filters against defamation:

https://twitter.com/DanielHolznagel/status/1512379315064913925?s=20&t=q8mT5paLDhMhfTuuPzJ6ww

Art. 8 + 9 (oders): no big surprises (still: these rules do not govern that providers have to follow the orders in itself)

9/X

Art. 10 + 11 (point of contact, legal representative) no surprises, Recital 36,36a: clarified: point need not be physical. Language problem not solved („providers … are encouraged to ensure that … languages chosen do not per se constitute an obstacle to communication“)

10/X

Art. 12 (ToS): 12(2) now requires to act (when applying ToS) with „due regard to … freedom and pluralism of the media“ … this seems innocent, but was probably lobbyied-in by publishing industry to slowly push „must-carry-must-pay“ claims

@Senficon

11/X

Art. 14 (notice and action …). Good: EP-proposals to make even stricter requirements (so providers will only have to act on „perfect“ notices) not adopted.

New Recital 40b: if host-provider cannot for technical reasons act upon notice, it should inform notice-sender...

12/X

Recital 41 clarifies: „Providers of hosting services should act upon notices in a timely manner,“ However, the big silence here remains: what action is required (evaluation-only or take-down? does only national law rule here? Can NetzDG stay in parts?

13/X

Art. 17 (Internal complaint-handling system): open for notice-senders too (as expected). Decisions to „restrict visibility“ are covered (council proposal). Problem: we still dont know what this is exactly and when „organic“ ranking decisions can fall under this...

14/X

Recital 42, 42a: „limiting accessibility“ as a content-moderation decision still not really clarified, what is it? shadow-banning as an example … but well, thats just an example.

15/X

Art. 18 (Out-of-court dispute settlement). Strange clarification in Art. 18: „The certified out-of-court dispute settlement body shall not have the power to impose the binding solution on the parties.“ That does not solve the problems ...

16/X

Of course the settlment-body cannot do it .. but can the DSC sanction a fine it the platform does not follow? Recital 44a clarifies that outcome of procedure is „not binding“. This must be taken by the words and should be reflected in Art. 18 itself.

17/X

No change: providers will still have to take part in the out-of-court proceedings. Still not solved: the unbalanced cost/reimbursment-scheme which imO will lead to abusive proceedings.

I think the whole procedure should be voluntary:

verfassungsblog.de/a-self-regulat…

18/X

Art. 20 (Measures and protection against misuse): no surprises, but EP-proposals (which would have weakened substantially) rejected.

Art. 24(1b) (target-ads-ban), surprise, it is not restriced to minors. This might be a mistake in the document?

19/X

Art. 24b (Online protection of minors) will come … however, we will see how strong this will be (remember: Art. 26,27 will outlaw meaningful minor-protection below 45 Mio users otherwise, e.g. existing youth protection, so Art. 24b crucial here).

20/X

Art. 26 (risk mitigation).

Interesting: Ad-hoc-risk assessments required „prior to deploying functionalities that are likely to have a critical impact“

Interesting: new risk mitigation on preventing deep-fakes and so on.

21/X

Recital 35: very nice: „The due diligence obligations are independent from the question of liability of intermediaries ...“ So no matter whether ECJ interprets Art. 7 narrowly, risk-mitigation would could circumvent and still allow for filter-obligations for VLOPs!

22/X

Art. 28 (Independent audit): lengthy provisions to secure independence of the auditor, but problem remains: Facebook & Co. will select and pay the auditor … I am sceptical and expect 1000pages documents being shy on platforms‘ shortcomings.

23/X

Article 33a Very large online search engines, Council position in.

Article 33b Supervisory fee is coming (as was reported) and – no surprise – is very complicated!

Article 73 (Evaluation): COM will have to evaluate the DSA now 3 years after entry into application (2025)

24/X

Thats it for me now.

Boiled down:

The DSA is strong on user-rights (too strong?), strong on transparency (good). Otherwise: lots of burocracy (too much), risk-mitigation in Art. 26,27 can theoretically be powerful, but I think Commission will have a hard lengthy job ...

25/X

... to do meaningful stuff here (will have to overcome whitewashing audits). Groundbraking in the DSA remains: bottleneck (safe-haven for platforms) Ireland looses competence für large platforms! Will the Commission do a better job?

26/26

@rochelag @prabprab @MCKettemann @RenateKuenast @ericgoldman @daphnehk @schrammmo @MarieTWeber @sveawindwehr @_blueck @benwagne_r @Anwalt_Jun @HateAid @LivaVikmane @hplehofer

@threadreaderapp unroll please

@JoanBarata @hutko @amelie_hldt @JPQuintais @schiebde @PLiesching @BRD_NK

following @rochelag comment here I think I was wrong on this conclusion! probably not convincing... 😔

clarification: tweet 22 I think I made a wrong conclusion... after rethinking, dont think this interpretation is strong.