The Toll Gate on an Emptying Road

Good evening, reader.

Editor's note: In May this newsletter published its analysis of the News Bargaining Incentive and proposed four amendments while the consultation window was open. In August, Parliament answered. This edition is the scorecard, and the honest tally is roughly one and a half out of four. The half-wins are real and worth naming. So is the loss, because the amendment that failed is the one that determines whether this instrument still collects anything by the end of the decade. Readers of last week's edition will recognise the shape: another wall left standing.

In today's Innovation Commons:

• The News Bargaining Incentive cleared Parliament on 20 August and commenced on 27 August, applying from the 2025-26 financial year: a 2.75 per cent charge on large platforms' Australian digital ad revenue, generous offsets for publisher deals, and a forecast $550 million in year one (Mi3; ATO).

• Scored against the four amendments we proposed in May: the diversity mechanics half-won (a 200 per cent offset uplift for smaller publishers, eligible expenditure spread across at least eight publisher groups), the 25 per cent single-recipient cap survived, use-of-funds reporting is nowhere in the reported final shape, and the AI exemption passed untouched (PIJI; RMIT).

• The arithmetic that follows: nearly one in ten Australians now reports getting news from tools like ChatGPT, Gemini or Perplexity. The levy's base is the channel news consumption is leaving. The exempt channel is the one it is entering.

• The falsification test: two numbers to watch across the first collection cycle, and one early tell.

The scorecard

In May, with the Charge Bill open for submissions, we mapped the closed loop: a well-designed levy attached to a distribution machine built to route the money to the same four organisations that captured the 2021 Code's pool and converted it into buybacks, acquisitions and roughly 450 redundancies. We proposed four amendments. Parliament has now answered, and honesty requires taking the wins and the losses in the same breath.

The wins first. The charge firmed from 2.25 to 2.75 per cent after Coalition pressure, and the final Act spreads eligible expenditure across at least eight publisher groups while lifting the small-publisher offset uplift to 200 per cent. That is not the 10 per cent cap and 30 per cent independent floor we asked for, but it is the same species of fix: the eight-group minimum roughly doubles the deal floor implied by the draft, and the uplift makes an independent masthead worth twice its face value to a platform's accountants. The communications minister's framing was blunt: platforms should deal directly with news organisations, and "if they do not, they will pay". Credit where owed. This is a materially better instrument than the April draft, and better than the 2021 Code it replaces.

Now the losses. The 25 per cent single-recipient cap survived, so a platform can still discharge the great bulk of its liability across a handful of majors. Nothing in the reported final shape adds the use-of-funds reporting that would have prevented the 2021 outcome, in which, by a former Nine chief executive's own admission, the money paid down debt and rewarded shareholders while the journalists left the buildings. And the exclusion we documented in May as clause 9 persists: the final legislation shows no plan to bring AI platforms inside the perimeter, an omission RMIT's analysts simply call the Act's AI gap.

The road, measured

Here is why the gap is not a detail. A University of Canberra study this year found nearly one in ten Australians already report using chatbots to get their news, and the number only moves one way. Ask an assistant what happened in the Budget and you receive a synthesised answer built from publisher reporting: no click-through, no impression, no attribution, and now, by statute, no levy. The charge applies to the Australian digital advertising revenue of designated search and social platforms, which is to say, to the mature channels news consumption is leaving. Every news question that migrates from a search box to a chatbot moves from the taxed column to the exempt one, shrinking the base while the untaxed channel compounds. Even the government's own projections ease after year one, from $550 million to $400 million; whatever mix of deal-uptake and base assumptions produced that curve, nobody involved is modelling the tolled road getting busier.

The wall this time

Readers of last week's edition will recognise the design fault, because it is the same one. The startup carve-out gates on turnover while the object it wants to encourage moves; the Incentive tolls links and feeds while the value moves to answers. The 2021 Code regulated the link economy just as feeds finished eating it. The 2026 Incentive regulates feeds and links just as answers begin eating them. On current form, the 2031 instrument will regulate answer engines just as whatever replaces them arrives. Five-year lag is not bad luck. It is what happens when instruments are designed around the incumbent channel's business model rather than around the value being extracted.

Three moves would finish the job, and all three were on the record before the Act passed. Bring AI answer services above a usage threshold inside the perimeter, the amendment independent publishers put into the consultation and we endorsed in May. Add use-of-funds reporting before the first payments flow, because the receipts from the last pool, $200 million a year in and 450 journalists out, are the strongest argument this newsletter has ever had to hand. And widen the eligibility register, which as of the May consultation listed 83 businesses against the regulator's own count of roughly 2,730 professional outlets in the country.

The test, so we can be scored the way we have just scored the Parliament: watch two numbers across the first collection cycle. One, what the Incentive actually moves, deals plus charge, against the $550 million forecast. Two, newsroom headcount at the recipients. If both hold up, this piece was too pessimistic, and happily so. If the money arrives and the journalists keep leaving, the country will have run the 2021 experiment twice, at a higher rate, with the fastest-growing channel watching from outside the gate. And one early tell: note which platforms choose to pay the charge rather than deal. A platform that prefers the toll to the traffic is telling you what it thinks the road is worth.

The toll gate in this week's illustration is staffed, the boom is down, and the road beyond it is grey. The traffic has not stopped. It has gone over the top, in small gold packets, out of the frame and out of the perimeter. The gate will collect from whoever is still on the road. The question the Act does not answer is what it collects when the road is empty.

— The Editor

P.S. Share this on LinkedIn/X to spark conversations with your colleagues.