hassleblad23

We had the same experience with our google ads across multiple campaigns. Tried everything from increasing the tCPA and daily limits significantly, to creating campaigns in 1st world countries only. Around 80% of the installs just opened the app and closed it after few seconds without performing a single action, not even a tap!

Google does not give you a detailed breakdown of where all the ad was shown within Google's services. They do give a breakdown of the 3rd party websites and apps where the ads ran and they are all terrible shorts drama and html5 games websites for me - some of them even had the same UI with different URLs! Shame.

Ideally I would like my app's ad to be placed in the Play Store only, but Google doesn't allow this control conviniently.

Next I am going to try using in-app actions for conversion instead of install, but I doubt it will be helpful at all.

Adding to this, Google's models can only be used with GCP while OpenAI's models can be used with Azure, Anthropic's models can be used with AWD Bedrock, in addition to their own platforms.

I'd love to see the Gemini models being available by other providers :) or if they just build a simple prepaid wallet like OpenAI and Anthropic.

That makes coasting easy for everyone.

Trying the iOS version and it looks pretty good so far. Good job! Maybe add an option to export the bookmarked posts?

I don't think the Enigma machine example applies here.

The nazi communications were decrypted by a highly centralized and secrative group, making it very difficult for the Nazis to figure out how they were doing it.

But in this case any vulnerability in the ransomware will have to be exploited by many of the affected people to decrypt their files, which means wide distribution, which means that a leak to the ransomware developers will happen sooner than later. If there is no wide distribution of the vulnerability, the ransomware developers win anyway.

You have to read my comment in context of the immediate parent which I replied to, not the OP.

The immediate parent comment says that if the vulnerability is publicly declared, attackers can easily patch it.

Paraphrasing my response: not publicly declaring the vulnerability is security by obscurity.. which does not work.

Don't attack a strawman.

I wonder at what point would the antivirus kick in. It doesn't require reading /dev/urandom for too long.

This is a game of cat and mouse, like it has always been. Cannot rely on security by obscurity.

Thats a great collection. Thanks for sharing.

Strobelight is open source as well.

This is a reason public college should be free.

I have noticed that asking an LLM to output a confidence score and the reason for assigning the confidence score, works really well. These are tangential to the actual task, but still improve the quality.

I wouldn't depend on the numerical value of the confidence score itself though. There is no way for the LLM to caliberate its confidence score wrt. multiple invocations on different data. I have found this metric to be mostly useless.

It works fine as a proxy to induce some thinking though.