You can declare partial victory though. Tech folks do now think more broadly about different constituencies. If a platform gets big enough, not "just" a company anymore. In the nick of time too, given rising nationalism and fall of the EU and other international institutions...
-
-
Replying to @balajis
Let me give an example. Tech issues were framed in a "what about individual privacy" narrative. That made little sense because the privacy/surveillance problem is a collective action/public goods problem. At the individual level, the trade-off is clear. Google maps, take my data!
1 reply 3 retweets 22 likes -
The question is not what happens if Google has my data, but it has a billion people's data, and what's their incentive structure? That barely got discussed. I've also advocated for privacy-preserving ML, federated learning etc. because there is no "click on yes/no" out of this.+
4 replies 5 retweets 31 likes -
A big challenge with the cryptographic solutions is just that the tech to use them easily "plug-and-play" style, doesn't really work yet. Anything more complex than encryption historically required a custom protocol, software tools that few understand, huge hits in efficiency....
8 replies 15 retweets 87 likes -
That said, this is changing. Zero knowledge proofs, for example, are rapidly commodifying, largely thanks to high demand from the blockchain/cryptocurrency space, and tools are quickly improving, eg:https://blog.iden3.io/introducing-circom-0-point-5.html …
2 replies 11 retweets 34 likes -
And I expect that over the next decade, MPC and maybe even obfuscation will follow suit. Development of these (largely public-good) technologies is hard, and was arguably largely subsidized by the 2017 crypto boom: one of the few clearly positive consequences of that event
1 reply 0 retweets 17 likes -
I personally am bullish on this stuff, but it needs to "just work", and it needs to be understandable to more than just 137 academics. Cryptographic solutions can only succeed when the *marginal cost* of using them is low, and that can only happen with high upfront investment.
1 reply 2 retweets 14 likes -
There's many incentive problems hindering adoption of better data architectures, and some will need political solutions (https://vitalik.ca/general/2019/05/09/control_as_liability.html … is one that's kinda already happening), but the problem of upfront cost of making the tech easy is one that we can move the needle on.
1 reply 0 retweets 6 likes -
Examples of these (once again public good) fixed costs: * Developing the algorithms * Optimizing the algorithms * Writing the software * Verifying the software's security * Education, making the algos and their properties understandable * Building ASICs for hardware optimization
1 reply 0 retweets 8 likes -
Replying to @VitalikButerin @balajis
https://www.wired.com/story/machines-shouldnt-have-to-spy-on-us-to-learn/ … I want this stuff to be as common as, say, brakes on cars or food safety.
1 reply 0 retweets 6 likes
People shouldn’t need to have chemistry labs in their basements to have food without lead in it. Same here with tech.
-
-
I'm 100% with you there.
0 replies 0 retweets 8 likesThanks. Twitter will use this to make your timeline better. UndoUndo
-
Loading seems to be taking a while.
Twitter may be over capacity or experiencing a momentary hiccup. Try again or visit Twitter Status for more information.