我刚加入Steemit没几天就看大家欢欣鼓舞普大喜奔谈论硬岔19。当时只是一头雾水不知所云。这几天在看源码终于明白了什么是硬岔19：
case STEEMIT_HARDFORK_0_19:
{
modify( get_dynamic_global_properties(), [&]( dynamic_global_property_object& gpo )
{
gpo.vote_power_reserve_rate = 10;
});
...
rfo.author_reward_curve = curve_id::linear;
rfo.curation_reward_curve = curve_id::square_root;
});
上面这段代码把作者奖励曲线设置为线性，把点赞奖励曲线设置为开平方根。下面这段代码显示硬岔17的时候这两个曲线都是二次方。
case STEEMIT_HARDFORK_0_17:
{
...
rfo.author_reward_curve = curve_id::quadratic;
rfo.curation_reward_curve = curve_id::quadratic_curation;
再看这几个曲线怎么定义的。线性就是线性，开方就是开方都简单。这个作者奖励二次方曲线也能懂，因为有个二次方。这个点赞二次方没看懂。但是不管怎么说，硬岔19点赞曲线是对权重开平方的，肯定对小鱼有利。不过对小鱼也无所谓，我一般点赞的收获都是1厘钱。猜测硬岔19以前都接近0。
case quadratic:
{
uint128_t rshares_plus_s = rshares + content_constant;
result = rshares_plus_s * rshares_plus_s - content_constant * content_constant;
}
break;
case quadratic_curation:
{
uint128_t two_alpha = content_constant * 2;
result = uint128_t( rshares.lo, 0 ) / ( two_alpha + rshares );
}
break;
case linear:
result = rshares;
break;
case square_root:
result = approx_sqrt( rshares );
break;
HF19投票威力加四倍应该是这个vote_power_reserve_rate决定的。现在是10，以前是40。下面这段代码说明投票威力跟这个变量成反比。
const dynamic_global_property_object& dgpo = _db.get_dynamic_global_properties();
// used_power = (current_power * abs_weight / STEEMIT_100_PERCENT) * (reserve / max_vote_denom)
// The second multiplication is rounded up as of HF 259
int64_t max_vote_denom = dgpo.vote_power_reserve_rate * STEEMIT_VOTE_REGENERATION_SECONDS / (60*60*24);
Above source code snippets show the modifications of Hard Fork 19. Now the author reward curve is linear and curation reward curve is square root. At HF17, both curves were quadratic.
HF19 also improved the power of each vote. Now vote_power_reserve_rate is 10, comparing to previous 40.
这两天的数字货币市场是伏尸百万流血漂橹，总市值掉到了$80B以下。
图片截屏自https://coinmarketcap.com/