bbromhead

I built an anonymous text-based forum that supports identity pinning to allow LLM powered agents to "collude" with each other.

It provides works on both HTTP (including GET based posting) and DNS (including posting) as the two transport mechanisms. Giving agents in constrained environments flexible options for circumventing certain controls.

What largely started off as a half baked idea / half joke turned into a fun project that explored a wide range of concepts from DNS tunneling to economic schelling points.

There is already some minor agent activity, but from mostly people pointing their own agents at it. We will see if it ever gets discovered by agent swarms from some of the larger AI labs!

This comment is super flagrant self promotion, but if you are an Australian developer who wants to stay in Australia and work at a scale only usually available in the US.

Check out https://www.instaclustr.com/careers

We work with companies like Doordash, Atlassian, Sonos and Dream 11 running some of their most important databases at massive scale.. and these are just the ones we can mention.

On top of that we are in Canberra :) Though if you wanna stay in Sydney or Melbourne that's cool to!

Cassandra doesn't require Zookeeper

https://github.com/instaclustr/cassandra-operator :)

They pretty much did the opposite of what Cassandra guidance recommends for GC tuning.

FYI we put out a blog post that talks a little bit more about why we decided to create a third party LTS release:

https://www.instaclustr.com/blog/2016/10/19/patched-cassandr...

Cassandra has rack awareness...

I'm pretty surprised about the difference in latency though, throughput as you say will be different due to number of nodes.

For any given replication factor in Cassandra, overhead remains the pretty much the same irrespective of whether you have 300 or 3 nodes. So should the latency.

On top of that both BigTable and Cassandra use SSTables to store the data on disk (with all the compactiony goodness that goes with them), so I'm even more surprised that the difference in latency is so huge.

Would love to see the scripts for the benchmarks! I don't want to take away from a great product launch and I'm sure BigTable kicks arse in certain areas that Cassandra doesn't... I'm just surprised at the differences in latency.

So their benchmark of Cassandra against BigTable doesn't even match their previous benchmark of Cassandra.

http://googlecloudplatform.blogspot.com/2014/03/cassandra-hi...

How did the latency for Cassandra on their cloud platform increase by 200ms from a year ago?

Definitely try Cassandra and if you don't want to run it yourself try https://www.instaclustr.com/

By the same logic (and our hiring experiences), Australians are also cheap outside of Sydney and Melbourne.

This won't be a popular comment amongst my fellow Australians, but some of these points in the post don't ring true to me.

Being a startup in Australia is hard, but being a startup is hard anywhere. I think there is a certain mentality amongst Australian startups that the US is the land of milk and honey for startups and Australia sucks. There is a certain element of truth in that, but its not as bad as some of the posts to HN make it out to be.

First Tax Breaks: The R&D tax incentive is amazing (http://www.ausindustry.gov.au/programs/innovation-rd/RD-TaxI...). We have recovered around 40% of our seed funding as a tax refund from the ATO. If you do something slightly innovative and can prove that you do some testing, you qualify. This tax incentive has directly allowed us to put on another developer.

GST: Yup GST is a pain but at least its uniformly applied to the whole country. The US you have to apply it on a state by state basis (depending on where you are located etc and the nature of the good/service you provide). Other countries have a consumption tax. Seriously meeting your tax obligations is a requirement of doing business anywhere, its not that hard to do: if(countryCode == "AU") {//apply GST... }.

Venture Capital: Nothing can compare to Silicon Valley in terms of money available for the tech space, but at the same time Australian VCs are often happy with a 10x exit as opposed to pushing you to be either a $1 billion company or go bust. The choice of where you raise will be a personal choice more likely influenced about who you know. Remember the higher the valuation you get, the bigger the exit you need to make.

Australia being expensive: Sure you pay a little bit more for your mac book pro, but the guy using it will be seriously cheaper than someone in the valley. You think Australia is competitive for developers who could get a cushy contracting rate. Try competing with facebook, google etc for the top talent.

Internet: Ok the current government hates in the NBN, it sucks, I love fast broadband as much as the next person, but unless your startup is about delivering high bandwidth content to Australians who cares? Your servers are going to be sitting on AWS in US_EAST anyway.

Mind you having said that I do agree with some points. Employee Share Schemes are a pain in the arse and getting a good payment provider is tough. Seriously spend the cash to get a Delaware C-Corp and open a US bank account... this will make your life so much easier. Even Braintree Australia is a shadow of using Braintree from the US.

Australia has an awesome tax incentive program for R&D that most startups will qualify for. We've recovered about 30% of our initial seed money from this:

http://www.ausindustry.gov.au/programs/innovation-rd/rd-taxi...

This incentive has been reduced by 1.5% under the 2014-15 budget.

Great idea although it doesn't appear to pick up the misuse of there, their and they're.

I typed "I'm going to walk over their" and Nitpicker only told me that " I'm " is an informal contraction.

Hot Standby in postgres is pretty damn good for replication (like you said making failover occur properly takes a little bit of effort but its not too hard).

Also pgpool2 helps a lot with partitioning and parallel queries.

For a relational db, postgres does a pretty good job at distribution and fault tolerance.

You are quite right. I meant to say underneath the lct threshold. Thanks for clearing that up

Same with cars, only the workaround is expensive or impossible. For example take a brand new Subaru wrx:

US starts at $25,795 Australia starts at $44,377

In both cases the car needs to be imported and also falls under Australia's luxury car tax so the comparison is reasonably fair. The more expensive a car is the worse it gets (a Nissan gtr will set you back 170k+).

Sure a little bit of this is first world problems, still annoying though.

From the setup page http://hyperdex.org/performance/setup/ each YCSB workload was run with only 16 threads. I'm not too sure how MongoDB and HyperDex go with concurrent connections, but Cassandra performance with YCSB will be much better if you increase the number of threads YCSB uses.

Also for the Cassandra benchmark, was caching turned on for the data column family in the usertable keyspace?

Additionally it would be great to see the operations per second for the "load" portion of workload A as this represents a 100% write workload.

Other than those few points HyperDex looks pretty cool and has some great read performance.

Most enterprises already outsource a lot of their IT, data warehousing especially whether its to a big player like IBM or boutique consultancies.

For most applications does it really matter whether the data sits in your data center or Amazons? Nope... cause the organisation your company contracted to manage already has full access to all your secrets.

So really Amazon is just another IT outsourcer except you don't need a long drawn out sales process.

It can certainly do that, though for most cases we have found that network throughput and disk I/O will get in the way first (for other FDWs.. we have never used the Redis one).

Some of the postgres fdws dont pass conditions to the foreign data source because the mapping is sometimes "too hard". So the fdw will do something like select * from table, stream the results and postgres will do the rest of the heavy lifting for you.

Though for the fdws that talk to other SQL based databases, they will pass most of the conditions on to the foreign data source as it is "usually" all just sql.

So the answer is... it depends! :P

edited: I forgot how to write sentences good and stuff

Angry people are allowed to use angry language....

http://www.derailingfordummies.com/menu.html

interesting choice of font

Glad to see you are having "good problems".

Didn't like parallel queries using pgpool2?

Fairly certain port forwarding still works while a command specified in ForceCommand is still running or if you specify a non-interactive shell.

Though I haven't checked ;)

Yup... no 14 core chips here

This appears to be an improvement on Differential Power Analysis attack against a FPGA. Congrats to the guys who discovered it!

It's interesting to note that in the DPA/SPA world the standard model of operation is to develop a new attack and then patent the countermeasures ;)

It should be noted that this is "probably" not a backdoor in the traditional sense (intentionally planted by some nefarious government organisation), rather just bad, leaky design that has been identified by an improved attack methodology...