This thread needs a bump.
There will eventually be ASICs for any mined algorithm which gains popularity. What this does is effectively act as a delay to help coins mature. It would take 8-12 months at least for asics to be developed for x11, maybe longer because it is dependent on economics. By that time those early adopter currencies with x11 will benefit from the asics, because ASICs benefit mature coins, and eventually the software technology would be developed for the next generation. There is a cycle to this.
Nothing is ASIC proof. Nothing. Resistant means it would take a new R&D effort to develop asics, and that literally can take several months. At which time a coins can mature who are early adopters of the tech.
There is definitely a life cycle here, and one we must recognize, and the life cycle is not entirely bad.
Currently. But I think there is a chance of using true randomness within an algo to make ASICs impossible to make. It just hasn't been done yet.
It is
impossible for a computer to create a truly random event.
Aw Hell, I'll put my two cents worth into the mix.
ANY predefined algorithm can be performed by an ASIC.
As long as the algorithm is known ahead of time, it can be performed by an ASIC. Some ASICs get around that rule by embedding specialized CPUs to adapt to minor changes in the logic while running, but they sacrifice speed in doing so.
I do many forms of engineering and FPGA programming is one of them. (For those who care FPGA stands for Field Programmable Gate Array. Clear as mud, right?) FPGAs are "firmware" versions of integrated circuits. They are, in a manner of speaking, reprogrammable ASICs. An FPGA can be reprogrammed to become most any digital circuit. One of the costs for this flexibility is speed. Very high-end FPGAs can come within 97% of the speed of an ASIC, but they are significantly more expensive at the high speed and large logic-quantity ranges. So, FPGAs are usually just used to prototype ASICs. Once the "firmware" version of the code is perfected on FPGAs, it is converted into photo masks for mass producing ASICs, analogous to a photocopier
A miner hardware vendor that is an exception to this is KnC Miner. They produce miners that are sold with the FPGAs in them rather than ASICs, for the most part. (They may now have some products with just ASICs in them or FPGAs and ASICs.) That is how they have gained speed to market, by skipping the ASIC chip production process. However, their pricing reflects the additional cost of using high-end FPGAs in their products.
To say an algorithm is ASIC resistant is just a pipe dream. There are only two things I can see that would make a cryptocurrency algorithm "ASIC resistant."
1) If the market for the coin is insignificant. Creating ASICs is a very time and capital consuming venture. (This is the crux of most of the constantly slipping ship dates for preordered miner hardware. Some aspect of the ASIC did not come out as predicted and adjustments must be made to the product design to accommodate those variances.) ASICs specialize in executing specific algorithms VERY fast. HOWEVER, the logic is cast in stone and metal, pretty literally. Once they are designed and set up for production, they can be churned out by the millions relatively cheaply. The only way to justify the creation of an ASIC is if there is sufficient market to consume those millions of chips to pay back the development costs... and then make a profit. Insufficient market, (as in the number of interested miner purchasers) no ASIC will be created. Period. Basic business economics.
The exception to this would be coins that, though different in some way, still use existing algorithms. Ie. A SHA256D ASIC can be used to mine any SHA256D based coin. The same is true for Scrypt ASICs. Even "Bitcoin-sCrypt" will subverted by this rule eventually. If there is enough money to be made, someone will figure out a way to get an advantage and turn a buck... er... coin.
2) If, as part of the design of the coin, the algorithm for the coin changes significantly and randomly during the life-cycle of the coin, it would be resistant to specialized hardware because the algorithms would not be fully known ahead of time. IMHO, this "shell game" method of implementing the logic would not work well in practice, but would lend itself to CPUs and GPUs because they can have their code changed "on the fly."
Those are the only two scenarios I see that would make a coin algorithm truly "ASIC resistant."
QFT, afa #2 Since it is impossible to create a random number on a computer and the coin is always open sauced, How could that be accomplished?
I have thought of another method but I'm not going to expose it in an open forum until the Idea has been fleshed out.