News for hackers. Stuff that matters.

senorcarbone

Re:OpenAI Launches Decisions API for Fast AI Choices on Wednesday September 30, 2026 @01:01AM (#49904536)
I think they simply use the LLMs softmax scores (uncalibrated confidence) so the answer is yes
Re:OpenAI Launches Decisions API for Fast AI Choices on Wednesday September 30, 2026 @01:00AM (#49904530)
From hallucinated output to hallucinating decisions :). The only way to automate decisions reliably is to do so without AI in the middle but just trustworthy statistics. Skeptical does this right https://skeptical.io
intelligence ≠ reliability.

Non statistical AI models such as LLMs, yes, even JEV (<3) are meant to hallucinate/approximate confidence so they are and always will be incapable of blind automation.

Instead, Skeptical turns any black-box AI into an error-conscious reliable pipeline with error cap. The technology has its roots on advanced statistical methods and external callibration.

Re:Integrating NVMe Disks in HopsFS (HDFS) on Tuesday October 16, 2018 @07:54AM (#18228559)
super cool. The world needs more success stories for NVM

Practical distributed systems MOOC for the masses

&quot;Reliable Distributed Algorithms&quot; is a distinct MOOC series that packs all the knowledge you need to design scalable and fault tolerant distributed services. It contains practical examples and exercises in Scala using the Kompics framework for property verification and also introduces the LASP language.Part 1 (starting Oct 17) and covers formal models, failure detectors, reliable...

( Read More... | no comments | 3 points | on HN )
Re:Announcing Apache Flink 1.0.0 on Thursday March 10, 2016 @03:05AM (#11257869)
Even that solution would approximate exactly-once delivery (eventually) but not strongly guarantee it.
Re:Announcing Apache Flink 1.0.0 on Wednesday March 09, 2016 @05:12PM (#11256092)
Flink guarantees exactly-once application state access/updates. There is no mention of exactly-once delivery anywhere and it is pretty much impossible to offer this for just any general sink. That does not exclude the possibility of offering a special-purpose transactional sink implementation in the future though (e.g. via external distributed locking/state reconciliation), just saying...
Re:Announcing Apache Flink 1.0.0 on Wednesday March 09, 2016 @04:44PM (#11255948)
Exactly-once delivery really depends where and how you actually "deliver". Even if your writes are idempontent you need to know if they have been properly committed on the other side, not that simple. If the system you are committing your output offers version control and/or proper transactional support then delivery can be eventually re-conciliated. Apache Flink's snapshotting algorithm solely guarantees exactly-once application state access, plain and simple.
Re:Thousands of Spaniards Leave Twitter for GNU Social on Saturday April 04, 2015 @02:31PM (#9321692)
I happened to work on some of the decentralised computing and networking projects in Barcelona. Guifi.net is the largest community network at the moment that among others lets users share and publish their content including internet access [1]. Last year I worked in Clommunity [2], an effort to build the first large scale community cloud. Guifi.net users can just install the Clommunity distro [3] on top of and use its service discovery and decentralised cloud management tools to contribute and benefit from marginal resources. I think it's a cool idea that we will keep revisiting in the long future...

[1] http://guifi.net/ [2] http://clommunity-project.eu/ [3] http://cloudy.community/