Its not the latest post in the argument but the one I want to address most, so:
The sha256 algo has become the workhorse of DigiByte I believe, so I think we will be keeping it for the time being. The insurmountable network speed of bitcoin is what many people believe provides its supreme security but with perhaps the imminent introduction of quantum machines that could change. It's not like satoshi had not thought of this, bitcoin was always destined to change with technology. after all that's the beauty of software. DigiByte strives to make this technology available to everyone and our multi-algo approach stops the big boys totally running away with the ball but we are still playing the same game, got a dog in the fight if you like!
edit: I'll think you will agree, that's better than a dog in your logo!
The SHA-256 miners (specialized highly concentrated miners) have a 4:1 advantage over GPU miners (those who really make up the widely distributed network) and are the main reason why price is kept so low since 50 SAT is their breakeven.
In theory, as well as in practice, the SHA-256 miners tend towards centralization, and, therefore, contribute not to security but greater network INsecurity and vulnerability as Jared has clearly explained. It's the small guy who helps promote decentralization, real network security, and widely accepted real world use . . . as has also been clearly explained on numerous occasions.
BTW, speed?
The trouble I see in your reasoning is that you are perfectly right in the smaller scopes of the area your argument is about, but it seems to fail to see the larger picture. Yes, SHA-256 tends towards centralization. Yes, a mining farm for BTC that contributes 'merely' about 1% towards BTC could in a matter of minutes provide 98 to 99% of the SHA-256 hashrate of DGB. A blockchain security nightmare for SHA-256 on its own. Not to speak of things like the power or influence Chinas government could have over the ASIC algos considering the combined hashrate of the mining farms and ASIC manufacturers based there, both accidentally or if they want/have reason to.
So SHA-256 is insecure and of no use? Better replace the ASIC algos with a CPU or GPU algo? Now, a few months back streaming based gaming got a bit of a news hype. With the CPU/GPU heavy calculations done in server farms anything capable to put a video stream on a screen and sent the controller inputs to the server would be able to play any AAA game in best resolution and quality. Now lets assume lower latency internet and other factors (hardware/electricity costs) makes this mainstream in near future. For example Steam with a low cost Steambox is really big in this. We are talking millions of gamers, hundreds of thousands to millions of GPUs in mostly unified gaming servers, running a unified OS controlled and serviced by remote access. Probably pretty good security against hijacking at least from outside, but still, has the thought that you still need at least somewhat big mining farms running 2 different types of ASICs before you could aim for breaking DGB a certain appeal now?
Additionally and less hypothetically, Jumbley mentioned the ASIC driven BTC hashrate surpasses the combined computing power of the top list of super computers. The same is said about the combined computing power of the server farms run by Google. Non-ASIC algos are usually still somewhat competitively mineable by CPU. Now taking over all of Googles servers should be quite impossible and why should Google combine all its resources to perform a hit on DGB? But like with BTC the same applies to DGB. Is it better to rely on 'They possibly could do it but why should they do it?' or aim for 'They can't do it even if they really want to'?
So my conclusion on the matter is: While SHA-256 on its own is indeed increasingly insecure it nonetheless hardens DGB against certain attack angles and in combination with the 4 other algos its making DGB one of, if not even the best, shielded coin/block chain. Is it really a good idea to risk this over profitability quarrels in between the algo/hardware types?