@eob@social.coop
@eob@social.coop avatar

eob

@eob@social.coop

Managing engineering team enhancing user-facing #privacy, fairness, and #AI compliance at #Google Search

Opinions here are strictly my own — I’m not speaking for Google

Formerly Bell Labs, HP Labs, and various startups

Have worked on chip design software, Internet collaboration software, IoT, computational aesthetics, search engine UI, and privacy

#EU citizen, from #Ireland

Have lived in Dublin, London, Princeton, San Francisco, and Calistoga

https://eamonn.org

This profile is from a federated server and may be incomplete. Browse more on the original instance.

eob, to random
@eob@social.coop avatar
Mer__edith, to random
@Mer__edith@mastodon.world avatar

Signal strongly opposes the newest #ChatControl proposal in Europe.

Let there be no doubt: we will leave the EU market rather than undermine our privacy guarantees.

This proposal--if passed and enforced against us--would require us to make this choice.

It's surveillance wine in safety bottles.

See more: https://www.patrick-breyer.de/en/majority-for-chat-control-possible-users-who-refuse-scanning-to-be-prevented-from-sharing-photos-and-links/ @echo_pbreyer

eob,
@eob@social.coop avatar

@Mer__edith @echo_pbreyer EU citizens, a timely reminder that you will be able to vote for your member of Parliament in about a week

This problematic law is coming from the Council (the upper house of the EU legislative branch representing the member governments)

It can be stopped by either the Parliament (the lower house of the EU legislative branch, representing the people) or the Commission (the EU executive branch, over whose composition the Parliament has some influence)

MEPs have power

aparrish, to random
@aparrish@friend.camp avatar

testing out a bit of code to display text on the gameboy by rendering 3px x 8px characters to VRAM tile data and tile maps, as needed at runtime. I like that the density of this allows more information on screen, but i'm worried that in practice it would be difficult to read—both because of the small size on an actual hardware screen, and because of the letters that don't take well to being rendered at 3px width, like Ms and Ws

eob,
@eob@social.coop avatar

@aparrish Are you restricted to 1 bit per pixel (on or off)?

If you could have multiple grayscale levels per pixel then you could use anti-aliasing to make the glyphs more readable

eob,
@eob@social.coop avatar

@aparrish Oh yes, now that I zoom in on the image I see the grey pixels.

Nice!

eob, to random
@eob@social.coop avatar

Powerful graphic for a story in the May 27 New Yorker about abortion, the courts, and the people

technicallymims, to random

feels like AI has done something to some tech journalists akin to what crypto did to some investors

RE: https://www.threads.net/@kylie.robison/post/C7PefkyvycB

eob,
@eob@social.coop avatar

@technicallymims Become irrational?

molly0xfff, to ai
@molly0xfff@hachyderm.io avatar

"it's all stored locally" is not a panacea for these alarming privacy-invading products!

what exactly is stored locally? what data is extracted from that local data and sent to the company's servers? is that local data being backed somewhere?

eob,
@eob@social.coop avatar

@molly0xfff Yes, you're right. Having personal data stored on your device is not a panacea. It replaces one threat model for another

It's going to vary for different people whether the threat of unwanted access to server data is worse than threat of their device being seized or subverted

Still, for the majority of people in most contexts, on-device personal data storage seems clearly the lesser privacy threat

(Even safer of course is to not participate, and avoid giving up any personal data)

eob, to random
@eob@social.coop avatar

As Ellen Ullman observed, we build computer systems the way we build our cities—over time, without a plan, on top of ruins

All software development is iterative

All software plans are wrong

The only way we advance is by learning from failures

carapace, to golang
@carapace@mastodon.social avatar

Is this a bug in Go's regexp package? This program should print false instead it prints true:

package main

import (
"fmt"
"regexp"
)

func main() {
r := regexp.MustCompile("^1|0$")
fmt.Printf("%v\n", r.MatchString("00"))
}

I checked against Python's re module and it works as expected:

>>> import re
>>> r = re.compile('^1|0$')
>>> print(r.match('00'))
None

eob,
@eob@social.coop avatar

@carapace Maybe your regex is being interpreted as

(^1)|(0$)

So it matches anything beginning with 1 or ending with 0

if you change it to

^(1|0)$

does that work as you expect to match either a single 0 or 1?

eob, to random
@eob@social.coop avatar

Well, it lasted less than a year

OpenAI had a much vaunted team working to make sure that AI would remain aligned with what humanity wants, even if the AI becomes smarter than humans

The leads and other team members have left, and Leike has expressed criticism about lack of support for this work

eob, to random
@eob@social.coop avatar

I remember an Ireland of poverty, clerical control, and stifling conformity

Today, the country took to the Eurovision stage, represented by a non-binary witch dressed in the trans flag

This is the New Ireland I celebrate / Éire Nua abú !

  • All
  • Subscribed
  • Moderated
  • Favorites
  • JUstTest
  • kavyap
  • DreamBathrooms
  • mdbf
  • InstantRegret
  • ngwrru68w68
  • magazineikmin
  • thenastyranch
  • Durango
  • rosin
  • Youngstown
  • slotface
  • khanakhh
  • tacticalgear
  • megavids
  • ethstaker
  • cubers
  • cisconetworking
  • normalnudes
  • modclub
  • everett
  • osvaldo12
  • GTA5RPClips
  • Leos
  • anitta
  • tester
  • provamag3
  • lostlight
  • All magazines