3 ms·
Self-responding - adding my position. I find most of these arguments against Google to be reasonable, inc complaints of monopoly. Personally I strongly mistrus
by WarOnPrivacy 6y ago
Self-responding - adding my position.
I find most of these arguments against Google to be reasonable, inc complaints of monopoly. Personally I strongly mistrust Google and have mostly divested myself from their offerings.
What I see Google did here was to vet the law's claims of fairness (Google gets money from our content so we should too) by stopping that arrangement. Once done, that particular unfairness was ended and the purported aim of the law was satisfied.
The grumpyness following Google's action confirms the law's intention is to create a new stream of tax revenue that AU Gov would grant (for now, anyway) exclusively to AU media entities.
I could potentially be behind a tax->news. Maybe. If it were constructed with more wisdom than this law is. I can not support this law because of it's shoddy construction (firehosing user data? srsly?) and because it's supplied purpose (ending unpaid usage of content) isn't truthful.
A larger issue may be that snippet laws seem to serve to further entrench Google's position by making news orgs dependent on Google for their operating revenue. This seems highly problematic.
It also grants new control over news funding to the AU Gov. If that control can be misused - to forward (AU Gov or 'partner' Gov) agendas - modern history tells us it will be.
Lastly, once this is normalized, who should pay next?
- judge2020 6y agoAnother boneheaded provision on the law is the requirement to give advance notice on upcoming changes to the algorithm to the news publishers. This is broken for both the reason of not being able to deploy 'algorithm changes' without delaying them 14 days (PDF page 23), which might not even be possible given machine learning's always-learning algorithm, and for the reason of news outlets for some reason being entitled to information on algorithm changes before everyone else in order to get a leg up in search results over anyone not in cahoots with the search partner. Law text: https://parlinfo.aph.gov.au/parlInfo/download/legislation/bills/r6652_aspassed/toc_pdf/20177b01.pdf;fileType=application%2Fpdf#search=%22legislation/bills/r6652_aspassed/0000%22 https://parlinfo.aph.gov.au/parlInfo/download/legislation/bi...
- astrange 6y agoSorry, what kind of machine learning is always-learning? That can be how regular programming works (using live statistics), but ML is all about very expensive offline training, the model doesn't change as you execute it.
- adonese 6y agoI think op might be referring to online learning [1]. [1]: https://en.m.wikipedia.org/wiki/Online_machine_learning https://en.m.wikipedia.org/wiki/Online_machine_learning
- astrange 6y agoRight, but all the methods in that article are old. Deep learning is typically not online at all, although you can do some things to personalize it.