"There is no sticking plaster for dirty and correctly modelled data".

One hell of a post in one hell of a thread.

Fabric Direct Lake Deep Dive YouTube video by SQLBits

Oh, this from @cwebbbi.bsky.social looks helpful. www.youtube.com/watch?v=ugLc...

Put another way, are data engineers even *qualified* to create the datasets in the gold layer in fabric? Maybe they shouldn't even have access to gold layer tables that will be consumed by #PowerBI because those tables are really owned by PBI modelers?

For instance, before Fabric, best practice was to use Views between #PowerBI and the DW. Now I take it, the equivalent of the transformative "Views" here is the Silver layer?

i.e. it is now more important than ever that folks populating the Gold layer in Fabric know exactly what PBI likes to eat?

Like, I used to do lots of renaming and shaping in #PowerBI because DBAs didn't really know (or possibly care) what PBI wanted.

But now those same DBAs are effectively charged with creating the semantic model in Fabric, and still might not really know/care what PBI wants.

Anyone got any #PowerBI resources concerning best practices when connecting PBU up to a Fabric datalake gold layer, for folks moving from Import to Fabric?

I take it that basically all the best practice rules and associated transformations I used to implement when importing just move up the chain?

What are all these faster dashboards supposed to do?

Who or what is gathering them up from the AI production line?

To what end?

The AI age is the "age of no consent" "Inevitability" means design decisions are no longer informed by user needs; now they are unilaterally imposed. The users are the ones being designed.

Users hate AI. So tech made it mandatory. Even if you don't use it, it pollutes what you read and how systems make decisions. The computer's hallucinated word is final.

And it needs all of your data to do it.

In the age of no consent, UX exists to normalized complete acquiescence to surveillance:

the thing i hate most about AI image gen is that no one does crude photoshops anymore, a tradition i will continue until the day i die

I'm sorry Dave, I really need you to agree to these new terms and conditions.

We still have #dataviz humour at least. bsky.app/profile/tang...

Those audible data labels are a novel idea!

Rare sighting of a column chart taking a dip at Hove Beach.

#dataviz

This part of it also sees it as full of extreme extremes.

I haven't been here for ages, is this still going? #dataviz people, are you there?

Maybe.*

*Maybe I’m still here. Maybe I’m still #dataviz. It’s hard to tell.

Jesus I had forgotten how hard it was to implement my Top N + Others pattern in a Small Multiple with timeseries in #PowerBI.

Just stupid amounts of code for something that should be implemented natively.

PowerBI will suck for as long as this is missing.

I already think that my industry (Business Intelligence, if you believe the marketing) is inefficient. I don’t think AI is gonna shift the dial for shareholders. Sure we will have more and faster dashboards but I think insights will not increase markedly and will be harder to find amongst the noise.

HELLLP SOMEBODY STOP ME RANTING.

a cartoon of homer simpson driving a green car ALT: a cartoon of homer simpson driving a green car

Our PowerBI tenant is already overwhelmed with these, when honestly a single chart created mindfully by an analyst chasing the right problem is worth more than the entire enterprise team gets paid in a year.

So while I - someone who has purchased more dataviz books that I have truthfully opened - can enjoy less friction creating dashboards coupled with the knowledge that I can tell AI to pull its design head in, it won’t be me putting out user-centric dashboards. It will be the data engineers.

I don’t doubt that building dashboards is amount to get a whole lot easier. I do doubt that insights are going to fall a lot thicker as a result. The few insights that leak from the few dashboards that produce them will probably be the needles in the haystack storage depot once this thing ramps up.

Oops Reassemble the analytics team! Reassemble the analytics team!

I think less friction creating dashboards just further amplifies “Management by dashboard” syndrome and “I can’t believe I get paid this much money to run a ringfenced dashboard fiefdom that costs a LOT” syndrome.

At the expense of this: benn.substack.com/p/oops

Disband the analytics team I mean, no, surely not. But maybe?

Im a Luddite. Here’s my manifesto:

open.substack.com/pub/benn/p/d...

Snap! Absolutely.

bsky.app/profile/insi...

Like, we already went through something similar when Power BI was released…it democratised many aspects of dashboard production, to the point where we now have “dashboard production as an outcome” mentality in practically every org that adopted it, at great cost but questionable shareholder value.

a woman in a white coat is standing in front of a door with the word so on her shirt . ALT: a woman in a white coat is standing in front of a door with the word so on her shirt .

I think the lack of friction encourages us to jump right into the tooling rather than jumping into the stakeholders shoes to make sure that the problem is worthy and that the problem is in fact the problem.

The other danger to me is of atrophy…not just of skills but of problem-solving approach and of stakeholder connection. I honestly think that slowing down and triaging underlying business problems given resources is much more important than speeding up solutions for every problem.