I believe that they are absolute values though. IIUC, a browser doubling their score on one of these benchmarks is an approximation for being twice as fast at performing the benchmarked functionality. Firefox's improvement was enormous in 2013, but these graphs make their improvement seem (eyeballing it) about 5 to 10x more than it was.
I agree, but it is rarely the fault of the person presenting them. Unfortunately most "easy to use" graphing technologies (with Excel being the worst offender) take the most deceptive graphing techniques and make them defaults.
They are deceptive because they exaggerate features of the data in order to improve perception of configuration changes. If every graph started at zero, many datasets would look like flat lines... not a great user experience when graphing.
Of course it's the fault of the person presenting them. If they are competent enough to understand what they are presenting, they are competent enough to adjust a y axis baseline.
