Hacker Newsnew | past | comments | ask | show | jobs | submit | alnorth's commentslogin

I'm a heavy Todoist user and I think it's great. I used to use org-mode, but all the Android apps I used for it were clunky and had issues with syncing when my file was concurrently edited somewhere else.

Todoist's API is pretty good too, so I've ended up building my own little webapp that fills some of the gaps in Todoist's functionality (e.g. finding a list of the projects that don't have a next action defined).


So if my competitor is using IngestAI and OpenAI use their data to train ChatGPT, could I literally just ask ChatGPT to tell me some secrets from my competitor's internal communication?


2023 is going to be very exciting times for security engineers.

It won't have the data, but it might have enough of an understanding of the data to leak important information.


While the model clearly can't retain all data, ChatGPT can regurgitate a lot of stuff verbatim.

Prompt:

> Recite the first two paragraphs of Neuromancer.

Response:

> Certainly! Here are the first two paragraphs of "Neuromancer" by William Gibson:

> "The sky above the port was the color of television, tuned to a dead channel.

> 'It's not like I'm using,' Case heard someone say, as he shouldered his way through the crowd around the door of the Chat. 'It's like my body's developed this massive drug deficiency.' It was a Sprawl voice and a Sprawl joke. The Chatsubo was a bar for professional expatriates; you could drink there for a week and never hear two words in Japanese."

(I have not checked how far you can get it to continue)

So perhaps it'll be a question of whether enough of your employees are feeding it copies of your data for it to retain it...


I bet that getting the right prompts won't be easy so it will probably fly under the radar and not immediately be detected. You can't search these weights with command-f. Fun times ahead...


Prompt engineering is only getting better. Also,

> You can't search these weights with command-f

Sometimes you can, https://clementneo.com/posts/2023/02/11/we-found-an-neuron


And good luck trying to add data to it without corrupting some other data it has encoded.


Does this problem disappear when using the Azure version of the service? If not, this is a pretty obvious market need: LLM + privacy.


Most AI companies won't want to offer that. They want to know if someone is using their service to instigate the next mass shooting or ethic genocide.


a good point. And what about companies that have on-premise storage?


yes, with OpenAI and also our type of apps security engineers have to move also move next level. And companies have to understand that it's context-aware only based on the knowledge-base you upload. It can not go and grab some data on your PC just because some one would ask it in chat))

BTW, Thanks for your comments! Appreciate it a lot.


This is a well known problem with this technology (although I haven't seen an official term for it, so we have been calling them "recovery attacks'). It's apparently the reason companies like Amazon have banned internal use of services like ChatGPT. I should add while it has been proven to occur the likelihood of something like this is very low. It's going to be a rare occurrence.


thanks for sharing. Do you think if all would be on the client's local server or cloud there would be still some, even rare, occurrence of that?


The problem could still occur but you would have to be capturing all the queries to your internal LLM systems and then using that data for training. You have complete control of the model so you could just choose not to do that and I would think data leaks of this nature would be less of a concern for an internal environment anyway. You would know that only authorized individuals would have access to the data. I suppose there could still be a very small chance of leaking data to unauthorized employees, but if a rogue employee wants to access data they should not have access to fishing an LLM would probably be the least productive way to do that. Your access logs for the LLM system would clearly display the attempts.

Some commercial services are starting to offer "Enterprise" licenses that prohibit the collection and use for training of your data and that would address the concern as well.


If a server was misconfigured OpenAI could have been trained on non public information. You can also poison OpenAI's dataset if you know it has been pulled by the service.


on a higher level of understanding, yes. But it would answer queries / it would be contextually intelligent only based on the previously uploaded knowledge-base. Did I get right your point?


The problem is that whoever you follow (whether they're friends, family or just interesting people) they ARE following the outrage parts.

And of course there's the "trending" sidebars, and ads and other stuff that gets injected into your timeline without you asking for it.

So things bleed in no matter how careful you are. The platforms are designed to drag you in and outrage you. Trying to avoid that is a constant battle - and you're going to give in to it from time to time.


Exactly this.

My current move is to delete Facebook completely (well, deleted that years ago). I just didn't find as much value there as elsewhere on the internet. And the psychological cost wasn't worth the little value there was there. So, complete deletion.

Twitter, I've deleted the mobile app and I follow no one. I have a separate list of profiles I find interesting. I have to physically go to that list and click on their profiles.

This keeps me out of 90% of the drama and into 90% of the worthwhile content.

It's hard to not look at "trending" but I'm trying. I wish there was a way to turn off "trending" and "timelines" completely.

End result: hugely more productive and psychologically lighter.


I use

twitter.com##section div[aria-label="Timeline: Trending now"]

In Ublock - origin rules to hide the trending sidebar


Better: https://github.com/insin/tweak-new-twitter

This will benefit from updates from the author whenever Twitter tries to foil blocking. It comes with a variety of tweaks like a separate timeline for retweets and forcing the latest timeline.


> Move retweets to a separate Retweets timeline, or hide them entirely

This alone completely changes Twitter for me. Making Twitter mostly original content? Sign me up.


Oh my gosh, that's beautiful. Thanks.


This is fantastic.


well thats handy

Thanks!


> Twitter, I've deleted the mobile app and I follow no one. I have a separate list of profiles I find interesting. I have to physically go to that list and click on their profiles.

Interestingly, this is the same solution I came to with Twitter, except I don't even keep a list of profiles, I just navigate based on memory and auto-complete in the address bar.


Cool. It's changed twitter completely for me.

My list is far too long to memorise, though.

And having it separate helps keep me honest (I think my memory would suffer biases) and out of my head.

Mainly though - the act of having to physically click a few times to get to the list gives me a moment of pause where I'm able to wake up and recognise whether I'm about to make good use of twitter or habitually jump into the time-worm-hole we all know.


I actually tried facebook for a month with other windows covering up the "trending" sidebar. It was a little better but the ads in the main area still annoyed me.

Then I started unfriending people that just posted stuff that annoyed me.

Then I realized after 3 days there was literally no new posts.

Then I just deleted facebook.

I don't miss it at all.

I've got multiple group chats with friends that I like. People post funny and interesting stuff there and we talk about it without having to deal with what someone's crazy uncle thinks about the topic.


This doesn't address all the issues you raise, but my hack to deal with the Twitter sidebar is to set it to a language and region I don't understand.

1. Click "Show more" at the bottom of "What's happening"

2. Click the "settings cog" at the top of the page

3. Uncheck "Show content in this location"

4. Click "Explore locations" and choose a location whose language you can't read and you know nothing about.

And to avoid outrage leaking to me via people who end up in my main feed I simply aggressively mute pretty much anyone who talks about politics or similar. Sure, I therefore lack their signal but avoiding their noise more than makes up for that.


I use

twitter.com##section div[aria-label="Timeline: Trending now"]

In Ublock - origin rules to hide the trending sidebar


Those platforms want your attention so they will show you more of what you “engage” (tricky buzzword) with, occasionally injecting a bit of noise to try to keep you from getting bored. This is in fact what many critics complain about yet like.

I just tuned FB to show me stuff I wanted by scrolling past stuff I didn’t, “liking” stuff I did/do, and clicking through to read/respond to comments. It’s mostly friend&family updates, apolitical jokes, animal pics and other anodyne stuff fun to see.

And I block ppl who are on the outrage train, typically just for the 30-day automatic cool down and FB gets the message.

It’s my attention and I don’t have to give it to people.


You can use a 3rd party client (I use Tweetbot) to go back to a chronologically ordered list of tweets, with no promoted tweets, no sidebar, you have to explicitly click to find trending topics.

You can also keywords (which I think you can do with the Twitter app as well) for more granular curation.

I think solutions like this allow one to benefit from Twitter without paying the heavy price that their shitty engagement tactics demand.


The platforms are designed to show you more of what you engage with. I can understand that for people who are addicted to outrage-bait, blocking might be the only way to break out of that, but if you engage with the parts you want to see more of and not with the parts you don't, facebook learns pretty quickly, IME.


I don't know that I agree with this, but even if we assume it's 100% true, this is still a problem given Facebook's addictive nature. Practically speaking, most people don't have the impulse control to pull themselves away from mindless, addictive content. And I don't think it's a huge surprise that a lot of the people who have gone down the Facebook rabbit hole are older. Less tech savvy, and perhaps not as sharp as they were at the height of their lives.


Every hobby could be seen as a mindless addiction. Heck, talking to friends in person is much the same - people fall into the same conversation patterns, have the same arguments and reminiscences over and over, miss them if they're not having them even if they don't really take pleasure in them at the time. At some point Facebook is just life.


That is so foreign. I know what you are talking about. But it's something I utterly despise. I don't get why people do it. Sure I understand people forget or mix up who they said what to from time to time. But the same thing over and over week after week just to 'talk'? Please just shut up and let me read a book or something.


That's why I don't do social media or follow other people's social media. If I want to connect with them texting, phone calls, zoom is much better instead of trying to get the same from online outrage machines.


Twitter provides some pretty decent tools for curating your feed


Xen only hit v1 in 2004 and was relatively niche.


You have to use the dropdown to select the right page. As they said, it's in the conclusion of the article.


I have very fond memories of foobar2000. Mostly I just loved how fast it was to use!

Does anyone know of an equivalent for the Mac?



While Deadbeef is a great player in general, it doesn't buffer the data when playing from a network drive, which causes a playback interruption every time there's a connection problem, even for a split second.

For this reason I've been resorting to using mpv as an audio player, even though it's interface is not really designed with music playback in mind.


QuodLibet is an excellent minimal yet customizable music player. It is a GTK app but the Mac version looks and works great.


Go ahead then.


It seems to be working fine for Spotify and Netflix. Is there something special about journalism that means it wouldn't work there?


It's barely working for Spotify and Netflix is testing ads.

Music and movies are highly repeatable and have production costs that users recognize and value. News and general content is not valued the same so nobody wants to pay for it. The severe dropoff in any paywalled site shows the effect of charging for news.


Who mentioned patents?


In the current US patent climate disaster it is not really a big leap to see something that is clearly not protected by copyright and then think "maybe they got a patent for it and throw licenses around because of that."


Say I wanted to learn COBOL, has anyone got any resources they'd recommend?


Programming in COBOL is best learned by signing up as a trainee employee in a company that has been using it for a long long time and that employs a bunch of old programmers who will be more than happy to teach you.

It's a quirky language, (omit a single '.' somewhere and you're in a for a world of hurt) has its neat parts and in general will come across as limiting once you have been exposed to other, more expressive languages. But that's probably also the reason it is still around, think of it as JAVA but then conceived in the gray past. The original intent was to make a programming language that would allow managers to program. That didn't quite work out.


Hey, apologies for randomly hijacking a comment for this question, but over the past years you've been popping up all over HN's comment section with (it seems) a rather eclectic and broad mix of experience, knowledge, and (possibly) talent. I'm almost starting to suspect you're multiple persons (but not really).

Have you ever written about how this came to be? I find myself to be similarly broad in interest/experience, but I'd judge myself much more superficial/limited in this regard because of the breadth of it all. I'm curious how you manage it all.


Hey There. Not really the place for this but I've had an 'interesting' life for want of a better description, I've written lots about it on my blog at http://jacquesmattheij.com/

If you want to take it off HN feel free: jacques@mattheij.com


You've never heard of Jacques Mattheij, inventor of the live webcam?


Nope, perhaps too young for that?

That said, the guy doesn't even have a Wikipedia page! So how should I know?


By being old, obviously.


MicroFocus have a Visual COBOL [1] that works in Visual Studio or Eclipse and targets the .net clr and the jvm. Trial downloads available. Or there is GNU COBOL [2] with docs and faqs [3].

I'm not a COBOL programmer, though, and not planning to be.

[1] https://www.microfocus.com/products/visual-cobol/

[2] https://en.wikipedia.org/wiki/GnuCOBOL

[3] https://open-cobol.sourceforge.io/


Depends on the country.

Are in the USA?

Edit: Clarification, you can't download a Mainframe VM so you need access to a Mainframe for training.


Programming in COBOL does not require a mainframe.


You're right.

I tend to think about what you need to learn to actually work in a IBM Mainframe environment which is greater than just learning the language.


I think you could use Hercules.

http://www.hercules-390.org


Guidelines | FAQ | Lists | API | Security | Legal | Apply to YC | Contact

Search: