Lets take a few hypothetical scenarios:
All ASIC's move from 28nm tech to 16nm tech.
-More work is being done, therefore more security
ASICBOOST is released for free and all ASIC's adopt it
-Same amount of work is being done, security is the same
ASICBOOST is patented and only specific miners can use it
-Same amount of work is being done, but causes miner centralization.
Bitcoin's security is provided by work (proof of work). Actual work has to be done to increase security. "Shortcuts" do not increase security. ASICBOOST doesn't do more work, it lets you pretend that you did more than you actually did. It is not an efficiency gain, it is a shortcut. It is disenguous to compare it to other efficiency gains where more work was done.
The correct terminology to describe ASICBOOST is that it is a cryptographic attack.
A cryptographic attack is a method for circumventing the security of a cryptographic system by finding a weakness in a code, cipher, cryptographic protocol or key management scheme.
The cryptographic attack used by ASICBOOST is colliding message blocks.
This same cryptographic attack, colliding message blocks, was used by Google in February 2017 to decrease the security of SHA-1 from 2128 to 261. This allows anyone with a powerful computer cluster to produce full hash collisions for SHA-1, completely breaking its security. This means that an attacker can produce two files with the same hash if they execute this attack and compute 261 operations.
More about the SHA-1 attack here:
This page contains two different files with the same SHA-1 hash proving that SHA-1 is not secure and cannot be used to verify the integrity of files.
Whitepaper on the colliding message block attack on SHA-1 that was used by Google:
ASICBOOST uses colliding message blocks to reduce the security of SHA-256 from 2256 to approximately 2255.48. In practice, this is negligible. However, if a new attack similar to ASICBOOST was revealed that reduced the security to somewhere in the order of 261, Bitcoin mining would be completely broken. It would be possible to mine a block, no matter the difficulty, with 261 operations, which is very achievable with today's technology.
Calling ASICBOOST an efficiency gain is very wrong.
Leaving cryptographic attacks unpatched sets a bad precedent that we don't care about these kinds of attacks. When a more serious cryptographic attack is found people will point to this one and say "why was that one allowed". It needs to be clear that we will patch any vulnerabilities on SHA-256