sarahjamielewis

sarahjamielewis, 7 months ago to random

There is lots of discussion about Electron / webp and - as someone who would never ever use electron for anything remotely approaching a security sensitive context - I do think it misses the mark.

Electron is bad because it shares an attack surface with the most attackable surface, but then extends it with all the functionality that was deliberately removed / never implemented because security.

(While giving developers very few tools to actually lock down that context in a meaningful way)

sarahjamielewis, 10 months ago to random

"The data you have given us is too valuable to let other people freely read it"

"Your web browser is rendering content in a way that we think is morally wrong"

It's just pathetic at this point.

sarahjamielewis, 1 year ago to random

There is an alternate timeline where the semantic web took off and there was wide investment in ontological tooling to ensure that the information in academic papers, websites, and applications was structured and accessible to future processing.

We instead live in a world where all the useful data is trapped inside proprietary formats, and entangled in meaningless prose - a world primed for large language models to come along and hallucinate the data that might contained therein.

sarahjamielewis, 1 month ago to random

Something that does trouble me is that most people who try out @cwtch try out the Android version - it is the way of the world that mobile computers are far more numerous than others.

But this does give a terrible first impression because as much as we have invested into Android over the years it still does not come close to the stability and usefulness of the desktop versions.

Metadata resistant communication is hard. Metadata resistant communication on mobile is harder.

sarahjamielewis, 3 months ago to random

Forever seeking a system/tool/technology/philosophy that fits between day-to-day getting things done, and long term goal planning.

Day-to-day I use a bullet journal for both task management and habit tracking - I've been doing it for years and it works great for any task or project whose state can be easily captured, and works as well for tracking long term progress of particular projects/goals.

But it's a terrible medium for e.g. managing research where the structure is far less defined.

sarahjamielewis, 2 days ago to random

"Note that Recall does not perform content moderation. It will not hide information such as passwords or financial account numbers."

The computer, however, will stop you from recording DRM'd content.

Find it fascinating that when faced with drawing safety and security boundaries, the primary beneficiary is not the owner of the device, or the person using it, but random corporations who control the intellectual property rights.

The system doesn't work for you.

sarahjamielewis, 6 months ago to random

A few thoughts on AI...

A collection of blobs on my computer no larger than dual layer blue ray can already flawlessly transcribe audio, generate any image I can conceive of (in countless variations and constraints), summarize and translate text and write mediocre code.

The open source tools for combining, controlling, and hacking these models together are gaining new features and capabilities every week.

The ethics of how these models came to be is important..but..that horse has bolted.

sarahjamielewis, 17 days ago to random

I'm somewhat perplexed by the new SecureDrop protocol - https://securedrop.org/news/introducing-securedrop-protocol/

Specifically: "The server is “untrusted” in the sense [it] learn[s] nothing about users & messages besides what is inherently observable from its pattern of requests, and it should not have access to sensitive metadata, or sender or receiver information"

Seems like a very weak definition of "untrusted", especially when two comparison techniques explicitly attempt to restrict knowledge derived from access patterns.

sarahjamielewis, 10 months ago to random

The one thing I am missing in life is a set of tools implementing a software development methodology that is somewhere in between agile-get-it-out-fast-and-iterate-trashfires and multi-million-dollar-bureaucratic-aerospace-programs.

A process that doesn't require numerous teams to implement - but formal enough to provide some resistance to a world happy to prioritize anything over assurance.

All I have so far are a set of loosely compatible informal procedures. Missing a unifying process.

sarahjamielewis, 4 months ago to random

A security/crypto meta-topic for my sanity:

1. In any context where it could possibly matter, cryptographic deniability doesn't hold any weight.

2. Any party trusted with delivering OS updates can (be coerced to) compromise that device/app.

3. The actual utility of properties like forward secrecy in a world with (2) depend on contorting adversaries into unrealistic shapes.

4. While useful, many have too much faith in honor-system security ("self-destructing messages" / "no screenshot flags")

sarahjamielewis, 1 month ago to random

I really, really don't want to be calling out specific people or projects, I don't think it's a useful thing to do - but it makes me so sad to see people, whose work I deeply respect, volunteering/writing/promoting a tool whose privacy claims are fundamentally unsound.

Privacy tools that a metadata resistant are essential, but please technically vet the projects you a promoting.

sarahjamielewis, 10 months ago to random

A few years ago I mostly stopped publicly publishing details of vulnerabilities I came across.

Sometimes I report them, but most of the time it isn't worth it - and often times it is clear it might even be costly, especially in recent years.

On top of that, because of the space I am in I constantly come across applications with risk models that are "not even wrong" - and I see people using these apps and it pains me.

Reminder: there is a huge gap between PR fluff & real world implementations.

sarahjamielewis, 3 months ago to random

It took me a long time but I finally understand that "python" isn't a language, "python" is a superposition of a dozen or so different languages.

For success with "python" you have to be ultra careful with ensuring that if the person who wrote the script used "python 3.9" that you also run it with "python 3.9" - if you don't you will be faced with hundreds of exceptions that have no relation to actual reality.

Never rely on distro packaging, always build from source. Use venvs liberally.

sarahjamielewis, 8 months ago to random

The reality on the ground is that as soon as the UK's Online Safety bill becomes law then the de-facto assumption must be that any service provider with significant exposure to the UK might be under a notice that mandates the compromise of the security and/or privacy of that service.

The statement made today - explicitly designed to defuse any tension that might have held up the bill - only re-enforces that position.

The framing that this is a "win" for online privacy is deeply disingenuous.

sarahjamielewis, 7 months ago to random

The AI safety discourse is identical to the encryption discourse resting on the fundamental assumption that some software (and by extension math) is so powerful that we must heavily restrict access to it.

The arguments are absurd on so many levels.

sarahjamielewis, 2 months ago to random

Every complicated system can be broken down into a set of solutions to problems that plagued an earlier, simpler system.

The only way to understand any system is to understand that sequence of problems.

Occasionally, in the process you discover one of those problems is no longer relevant - requirements and environments change over time.

Sometimes, you find one adjustment supersedes another without removing the resulting complication.

The monument, as it stands, rarely reveals its purpose.

sarahjamielewis, 2 days ago to random

Software request: I'm looking for a tool I can use to manipulate nodes in a graph. Specifically I would like to be able to:

• Add new nodes to the graph (not a tree)
• Create multiple distinct edge relationships between nodes (bonus if the tool lets me formalize these edge types)
• Have nodes contain notes, perhaps be typed
• Export the graph to a reasonable (text) file format for external processing
• Explicitly not an image editor or diagram tool.
• Run on linux / be open source (flexible)

sarahjamielewis, 4 months ago to privacy

2023 was somewhat of a Red Queen's Race for #privacy, #p2p, #decentralization

At the end of it I feel like we had to run much faster just to stay in the same place.

Much like the rest of the space, funding for @openprivacy and @cwtch took a hit, and we have to continue to squeeze ever more out of the amazing support we do get.

When I look at where Cwtch is now compared to a few years ago I couldn't be prouder.

So much more to do, so many better worlds to build.

sarahjamielewis, 6 months ago to random

There are so many problems with the UK's Online Safety Act. But reading through the Ofcom consultation document I'm struck by a few things.

• Every implementation cost cited is about 2-3 orders of magnitude smaller than I would have come up with if asked for a conservative estimate for the kinds of services cited.

• The "Defences" section which explicitly carves out a defence for harassment of queer people - really underlines the type of "safety" this act provides.

sarahjamielewis, 9 months ago to random

Since people asked, some opinions about the Rust ecosystem:

0. I like the language. It's my default.
1. So much dependency bloat. Almost as bad as Python. Almost.
2. A general feeling that unmaintained libs will slowly cease to compile. (There was a time when many useful features in libs were nightly only...and the unmaintained ones don't build anymore)
3. I've built some personal tooling for Rust and the sheer complexity of some introspection components makes me weary of ever trusting it.

sarahjamielewis, 3 months ago to random

Really uncomfortable with (otherwise cool) organizations using the presence of cryptography to back up a security/privacy claim that is 100% policy based.

Just because they don't do a thing doesn't mean they can't do a thing.

"We don't know who you talk to" (because we don't log that information as it passes through our servers)

is a very different claim than...

"We don't know who you talk to" (because we physically and computationally will never have access to that information)

sarahjamielewis, 1 month ago to random

I would really appreciate, and be willing to pay for, a news source that restricted itself to covering legislative, judicial, and corporate machinations at the local/regional/national level while staying away from reporting on press conferences / inane social media statements / speculation / punditry.

i.e. reports on what people are doing, rather than what they are saying.

Would appreciate recommendations along these lines.

sarahjamielewis, 9 months ago to random

Experience has shown me that there is no real way to combat "not even wrong" claims about privacy and security in the secure communications space.

Demonstrating critical issues results in hostility and a quick patch that does nothing to fix the underlying systemic issue (at best).

Yes I find myself growing tired of holding my tongue while these apps are promoted or, somewhat more dispiriting, held up as models of good privacy engineering.

Caveat Emptor?

sarahjamielewis, 29 days ago to random

People have a right to access and use secure tooling that enables them to leverage modern cryptography.

The alternative is absurd. A demand to deliberately subvert foundational economic infrastructure. A position that should be laughed out of any sensible room.

sarahjamielewis, 1 month ago to random

A topic I would love to read a deep analysis on is how certain actions e.g. blocking, moderation/filtering, "self-deleting" messages etc. transform from passive server-side actions to client active actions in decentralized systems and if/how that breaks down against existing ingrained metaphors and expectations.