So the city of exquisite high paying banks focusing on ultra high net worth individuals is becoming to expensive for those very banks.
And robotics is really here to take over control.
I just wonder what that means for the notoriously tight housing market.
And robotics is really here to take over control.
I just wonder what that means for the notoriously tight housing market.
0
0
0
23
Most investment cases name their risks. Far fewer name the observation that would end the investment.
The difference is not semantic. A risk is something that might happen. A stop condition is a decision rule:
a specific, observable event that, if it occurs, ends or redirects requiring a new argument. Risks get managed. Stop conditions get triggered.
They are rare for structural rather than intellectual reasons. Naming the condition that would kill a proposal, inside the document that asks for approval of that proposal, reads as arguing against yourself.
The incentive runs the other way, and it runs that way for competent and honest people.
Two things make it easier.
First, write it before approval rather than after. Once capital is committed the same sentence changes meaning: it stops being a criterion and starts being an accusation. Stage-Gate practice puts explicit go, kill, hold and recycle criteria at each gate for this reason: the decision rule is agreed while all the outcomes are still open.
Second, separate writing the condition from owning the outcome. In cross-functional teams it’s not uncommon to find different functions hold different standards of proof, and that difference is not irrationality. That diffenrces is the job.
Someone whose performance does not depend on the project continuing will write a cleaner condition than the sponsor, and will still believe it a year later.
Gary Klein's premortem is the cheapest version: assume the project has already failed, then ask the room to say why. It works because it converts a disloyal question into an assigned task.
A usable stop condition has four parts:
- the observation, stated so that two people would agree it had happened;
- the date or milestone at which it is checked;
- the person who checks it, and who does not carry the outcome;
- what follows on the trigger: stop, resize, or hand to a different route?
Anything missing one of the four is a sentiment rather than a condition.
The failures that get remembered are rarely failures of foresight. In most of them somebody named the condition, on time, in writing. It was read as pessimism, or as a threat to a plan that had already acquired its own momentum, and the ride continued.
A crossroads is not a place of despair. It is the last point at which turning is still cheap.
Who is allowed to write the stop condition, and do they own the outcome?
Image: Woman Holding a Balance, c. 1664, Johannes Vermeer.
The difference is not semantic. A risk is something that might happen. A stop condition is a decision rule:
a specific, observable event that, if it occurs, ends or redirects requiring a new argument. Risks get managed. Stop conditions get triggered.
They are rare for structural rather than intellectual reasons. Naming the condition that would kill a proposal, inside the document that asks for approval of that proposal, reads as arguing against yourself.
The incentive runs the other way, and it runs that way for competent and honest people.
Two things make it easier.
First, write it before approval rather than after. Once capital is committed the same sentence changes meaning: it stops being a criterion and starts being an accusation. Stage-Gate practice puts explicit go, kill, hold and recycle criteria at each gate for this reason: the decision rule is agreed while all the outcomes are still open.
Second, separate writing the condition from owning the outcome. In cross-functional teams it’s not uncommon to find different functions hold different standards of proof, and that difference is not irrationality. That diffenrces is the job.
Someone whose performance does not depend on the project continuing will write a cleaner condition than the sponsor, and will still believe it a year later.
Gary Klein's premortem is the cheapest version: assume the project has already failed, then ask the room to say why. It works because it converts a disloyal question into an assigned task.
A usable stop condition has four parts:
- the observation, stated so that two people would agree it had happened;
- the date or milestone at which it is checked;
- the person who checks it, and who does not carry the outcome;
- what follows on the trigger: stop, resize, or hand to a different route?
Anything missing one of the four is a sentiment rather than a condition.
The failures that get remembered are rarely failures of foresight. In most of them somebody named the condition, on time, in writing. It was read as pessimism, or as a threat to a plan that had already acquired its own momentum, and the ride continued.
A crossroads is not a place of despair. It is the last point at which turning is still cheap.
Who is allowed to write the stop condition, and do they own the outcome?
Image: Woman Holding a Balance, c. 1664, Johannes Vermeer.
2
0
0
35
Seeing a lot of labor/wage arbitrage in 2026.
Looks like it sits somewhere between
- operational excellence aka cost cutting,
- economic-aid aka sending money to lower wage / developing countries,
- ...
Pretty confident there are higher order effects that are being neglected by many. Best of luck to you.
Looks like it sits somewhere between
- operational excellence aka cost cutting,
- economic-aid aka sending money to lower wage / developing countries,
- ...
Pretty confident there are higher order effects that are being neglected by many. Best of luck to you.
0
0
0
28
The miracle of an interdependent economy machine creates is frankly unimaginable by so many.
x.com/macyagilliam/s…
x.com/macyagilliam/s…
I would actually be curious from anti data center people on whether they think the Gwangyang Steel Works factory on South Korea should have been allowed to be built. What general rules do they follow for deciding when a large new industrial building should be allowed?
0
0
0
46
Heterogeneous integration across the scales. Integrated industrial complex’ and integrated (heterogenous) circuits share a few concepts.
Let’s dive into metaphors for a moment.
Photons are like rail cars, they can’t get around very tight corners, it takes quite some extra effort to load and unload them, the further you can integrate them into your infrastructure, the more effective and efficient they become, they can carry vast amounts on thin lines — so businesses go to great lengths to utilize them businesses.
And analogous: electrons are like automobiles, they just need general pavement, which also is much easier to maintain, on that pavement they can get to where they are needed not ultra fast but with a good enough combination of speed and flexibility.
Let’s dive into metaphors for a moment.
Photons are like rail cars, they can’t get around very tight corners, it takes quite some extra effort to load and unload them, the further you can integrate them into your infrastructure, the more effective and efficient they become, they can carry vast amounts on thin lines — so businesses go to great lengths to utilize them businesses.
And analogous: electrons are like automobiles, they just need general pavement, which also is much easier to maintain, on that pavement they can get to where they are needed not ultra fast but with a good enough combination of speed and flexibility.
I would actually be curious from anti data center people on whether they think the Gwangyang Steel Works factory on South Korea should have been allowed to be built. What general rules do they follow for deciding when a large new industrial building should be allowed?
0
0
0
44
It’s all helicopters!
Choppers are back!
Choppers are back!
>designing a new heavy lift platform
>ask DARPA if it can be weird or just helicopter
>they don't understand
>pull out illustrated diagram explaining what is helicopter and what is weird
>they laugh and say "haha just design a good one"
>compete
>its all helicopter
0
0
0
54
Jona Engel retweeted
If DefendTex can pass the DARPA course, they will likely win
Their drone is 12lbs with a 112 lb payload weight (9.63:1)
The top teams don't want to crash out of competition before graded runs, so there are a lot of test flights
Their drone is 12lbs with a 112 lb payload weight (9.63:1)
The top teams don't want to crash out of competition before graded runs, so there are a lot of test flights
60
107
3,669
480,318
Big acquisition announcement day today. Industrial and materials consolidation in full swing. Expecting more to come.
0
0
0
25
Before running a test, write down which decision it is allowed to change.
That one sentence separates evidence from activity.
In industrial technology, it is easy to accumulate proof without reducing the uncertainty that blocks the next commitment. More microscopy. Another urvey. A larger report. All may be technically sound, yet the investment decision remains exactly where it was.
A useful proof ladder starts from the decision and works backward:
1. What decision is currently blocked?
2. Which uncertainty is actually blocking it?
3. What is the smallest observation that would materially change our belief?
4. What threshold separates the possible outcomes?
5. What will we do in each outcome?
The last question is the one most often missing.
If a positive result leads to more testing, a negative result leads to more testing, and an ambiguous result also leads to more testing, then the experiment has no decision rights. It may create knowledge, but it is not yet controlling investment.
Stage-gate systems formalize this idea with explicit decisions such as proceed, stop, hold or repeat. The method is not perfect, but the principle is valuable: evidence earns its place by changing what happens next.
The honest counterpoint is that not all research should be forced into an immediate business decision.
Exploratory science often matters precisely because its value cannot be specified in advance. Early discovery needs room to surprise.
But once a team says it is reducing market or development risk, the standard changes. The test should be tied to a named uncertainty, a threshold and a consequence.
The strongest experiment is not necessarily the most sophisticated one.
It is the smallest credible proof that can make the next decision different.
That one sentence separates evidence from activity.
In industrial technology, it is easy to accumulate proof without reducing the uncertainty that blocks the next commitment. More microscopy. Another urvey. A larger report. All may be technically sound, yet the investment decision remains exactly where it was.
A useful proof ladder starts from the decision and works backward:
1. What decision is currently blocked?
2. Which uncertainty is actually blocking it?
3. What is the smallest observation that would materially change our belief?
4. What threshold separates the possible outcomes?
5. What will we do in each outcome?
The last question is the one most often missing.
If a positive result leads to more testing, a negative result leads to more testing, and an ambiguous result also leads to more testing, then the experiment has no decision rights. It may create knowledge, but it is not yet controlling investment.
Stage-gate systems formalize this idea with explicit decisions such as proceed, stop, hold or repeat. The method is not perfect, but the principle is valuable: evidence earns its place by changing what happens next.
The honest counterpoint is that not all research should be forced into an immediate business decision.
Exploratory science often matters precisely because its value cannot be specified in advance. Early discovery needs room to surprise.
But once a team says it is reducing market or development risk, the standard changes. The test should be tied to a named uncertainty, a threshold and a consequence.
The strongest experiment is not necessarily the most sophisticated one.
It is the smallest credible proof that can make the next decision different.
0
0
0
32