There are managerial challenges when you are responsible for for is working remotely. For the first 3 years or so of Hypernumbers we were rotating between 7- and 6- day working, going down to 5-day working in the summer.
The real concern when you are working with young developers (I was mid-40s, Dale was early 20s) is about burnout. Particularly with software developers. Young men working long and engaging hours with addictive work practices, with the usual poor dietary habits - this is a bad accident waiting to happen. At if.com we had a particular developer who physically turned green during a big push. Easy to spot in the work place, remotely not so much...
Dale has a quite common tendency for diurnal drift - a 25 hour cycle. Nothing wrong with it, slightly alarming when you sign off in chat at midnight, go to bed, start up at 7am when you get up and he's still there. Diurnal drift can rapidly turn into social isolation though (and a one way ticket to the bin).
Our industry has a lot of people at one end of the autistic spectrum - and remote working is not an policy for everyone.
Domestic circumstances also count a lot - support at home, friends, community stuff.
I heartily agree with what Dale says about communication, practices and procedures. I have long thought that the main lesson of open source projects is that they can only succeed if all the modalities are tip-top. We adopted 'open source processes' at hypernumbers to drive quality - and remotish working helped drive them in.
The circumstances of hypernumbers were a bit different though. Normally you would look to colocate the person specifying the system and the people who are writing it. With hypernumbers we spent the first couple of years essentially cloning Microsoft Excel so the business processes were optimised for remote working. (This is also one reason why open source programmes are dominated by clones.)
So it is a bit more nuanced I think than in his write up.
Would I hire remote workers again? Yes I would. Contrary to received wisdom I try and hire people from the 'other' end of the autistic spectrum (the end I would put Dale on).
On conventional measures of productivity (production SLOC per person per day) I would regard the experience at hypernumbers very highly - although the total difference in working procedures versus if.com where I have 'before' measurements from make it hard to allocate outcomes to a particular aspect of how we worked.