Most antibodies to viruses don't last forever, the question is not if you can get COVID19 twice (you almost surely can) but how long in-between are you protected.
And the evidence is unclear, and in some cases worrying:
As @tpnCC and @seppin said, there is significant clinical & scientific uncertainty about covid19 antibody response and duration, partly because there hasn't been enough time to collect the relevant data (for covid19) and partly because of prior resarch indicating reduced human antibody response to SARS.
Lower in this thread I commented about a 2011 resarch paper that demonstrated anti-SARS antibody levels fell by half 2 years postinfection and were undetectable 6 years postinfection in 21 of 23 SARS patients tested.
There is a big problem with optimistically assuming immune systems will respond like SARS even, because we know in general terms the more severe the symptoms the longer the immunity lasts, but wait ... huge huge numbers of people with Covid-19 are asymptotic, that is not the cases with it’s SARS and MERS relatives! wake up people, herd immunity is delusional and dangerous. more likely is very short immunity, just like the other 4 Corona family virus which cause many of our common colds, high asymptotic rates, and reinfect us very very frequently and NOT due to mutations!
If those SARS numbers are in the same vein as what is true for COVID19, the only answer is a vaccine. Otherwise our current state of affairs is the new normal for the foreseeable future.
It's estimated that ~14% of common cold cases are caused by coronaviruses (HCoV-OC43, HCoV-229E, HCoV-NL63)... so although it seems particularly difficult, if we do manage to discover how to develop effective vaccines against Covid19, perhaps the silver lining is that vaccines against the related viruses responsible for a significant fraction of the common cold will be in reach.
In that case COVID19 will play out like the 1918 flu, it will hit in waves that only mass quarantine will help mitigate, until enough immunity is reached. Likely 18 months - 2 years.
Just to spell it out: herd immunity depends on sustained immunity postinfection. It seems possible, perhaps probable, that covid19 "immunity" may have an unusually short duration or attenuated response, resulting in an effective herd population that is a fraction of the theoretical herd population. I.e. if 60% of Americans have been infected and recovered after 12 months, but antibody levels drop 25% after 12 months, the true number of immune americans comprising the herd will be less than 60%.
If so, we may need to consider pursuing a relatively slow and steady relaxation of lockdown to tune the rate of new infections such that herd immunity is kept as stable as possible - i.e. flatten curve into truly flat-but-nonzero line
Lower in this thread I commented about a 2011 resarch paper that demonstrated anti-SARS antibody levels fell by half 2 years postinfection and were undetectable 6 years postinfection in 21 of 23 SARS patients tested.
see this chart https://www.jimmunol.org/content/186/12/7264.long#sec-7